Timeline

Jun 16, 2022:

11:58 PM Changeset in webkit [295630] by timothy_horton@apple.com

• 10 edits
  8 adds in trunk

Record InteractionRegions per RenderLayer, instead of all on the root
​https://bugs.webkit.org/show_bug.cgi?id=241503
<rdar://problem/93855866>

Reviewed by Simon Fraser.

• Source/WebCore/page/DebugPageOverlays.cpp:

(WebCore::pathsForRegion):
(WebCore::InteractionRegionOverlay::activeLayer const):
(WebCore::InteractionRegionOverlay::activeRegion const):
(WebCore::InteractionRegionOverlay::drawRect):
Hit test to the correct layer and retrieve regions from that layer for the debug overlay.
Also, fix some coordinate conversion now that we actually see non-root layers.

• Source/WebCore/page/InteractionRegion.cpp:

(WebCore::cursorTypeForElement):
(WebCore::interactionRegionForRenderedRegion):
(WebCore::absoluteBoundingRectForRange): Deleted.
(WebCore::regionForElement): Deleted.
(WebCore::interactionRegions): Deleted.

• Source/WebCore/page/InteractionRegion.h:

Refactor InteractionRegion to take the same arguments as EventRegion::unite(),
so we can call it from there. Use the painting-originated rects in the Region,
instead of computing them ourselves.

• Source/WebCore/rendering/EventRegion.cpp:

(WebCore::EventRegionContext::unite):
(WebCore::EventRegionContext::uniteInteractionRegions):
(WebCore::EventRegionContext::copyInteractionRegionsToEventRegion):
(WebCore::EventRegion::computeInteractionRegions): Deleted.

• Source/WebCore/rendering/EventRegion.h:

Collect InteractionRegions in a HashMap by element identifier, in order to
unite the rects for a given element -- but not *across* elements, like other EventRegions.
We need to keep them separate so that we can add an indicator for the united region.
Also, unusually, we maintain this map on the EventRegionContext as we paint, and copy them
to the serialized type at the end, to avoid having two different members on
EventRegion that are only valid on opposite sides of the process boundary.

• Source/WebCore/rendering/RenderLayerBacking.cpp:

(WebCore::RenderLayerBacking::updateEventRegion):
Stop computing interaction regions for the root layer.

• Source/WebCore/rendering/RenderBlock.cpp:

(WebCore::RenderBlock::paintObject):

• Source/WebCore/rendering/RenderLayerCompositor.cpp:

(WebCore::RenderLayerCompositor::updateScrollLayerClipping):

• Source/WebCore/rendering/RenderReplaced.cpp:

(WebCore::RenderReplaced::paint):

• Source/WebCore/rendering/TextBoxPainter.cpp:

(WebCore::TextBoxPainter<TextBoxPath>::paint):
Plumb the renderer through to EventRegionContext so that we can use it in
InteractionRegion::interactionRegionForRenderedRegion.

• LayoutTests/interaction-region/click-handler-in-shadowed-layer-expected.txt: Added.
• LayoutTests/interaction-region/click-handler-in-shadowed-layer.html: Added.
• LayoutTests/interaction-region/inline-link-in-layer-expected.txt: Added.
• LayoutTests/interaction-region/inline-link-in-layer.html: Added.
• LayoutTests/interaction-region/inline-link-in-composited-iframe-expected.txt: Added.
• LayoutTests/interaction-region/inline-link-in-composited-iframe.html: Added.
• LayoutTests/interaction-region/inline-link-in-non-composited-iframe-expected.txt: Added.
• LayoutTests/interaction-region/inline-link-in-non-composited-iframe.html: Added.

Add some tests.

Canonical link: ​https://commits.webkit.org/251635@main

11:23 PM Changeset in webkit [295629] by commit-queue@webkit.org

• 2 edits in trunk/Source/WebCore/html/canvas

Fix InspectorScopedShaderProgramHighlight for indexed blend state
​https://bugs.webkit.org/show_bug.cgi?id=241463

Patch by Alexey Knyazev <​3479527+lexaknyazev@users.noreply.github.com> on 2022-06-16
Reviewed by Kimmo Kinnunen.

• The program highlight helper now correctly restores

indexed blend state when OES_draw_buffers_indexed is enabled.

• The program highlight helper now skips non-default FBOs.
• Optimized execution to directly call into GL context

avoiding extra validation and value unboxing.

• Source/WebCore/html/canvas/WebGLRenderingContextBase.cpp:

(WebCore::InspectorScopedShaderProgramHighlight::showHighlight):
(WebCore::InspectorScopedShaderProgramHighlight::hideHighlight):
(WebCore::InspectorScopedShaderProgramHighlight::saveBlendValue): Deleted.
(WebCore::InspectorScopedShaderProgramHighlight::hasBufferBinding): Deleted.
(WebCore::InspectorScopedShaderProgramHighlight::hasFramebufferParameterAttachment): Deleted.

• Source/WebCore/html/canvas/WebGLRenderingContextBase.h:

Canonical link: ​https://commits.webkit.org/251634@main

11:05 PM Changeset in webkit [295628] by Wenson Hsieh

• 4 edits in trunk/Source/WebCore

[iOS] Callout bar is sometimes obscured when selecting Live Text in a fullscreen video in AVPlayerViewController
​https://bugs.webkit.org/show_bug.cgi?id=241700
rdar://95310525

Reviewed by Tim Horton.

Even after the changes in r295052, UITextEffectsWindow-hosted text selection views (e.g. grabber
dots and callout bar) are still _sometimes_ obscured behind the fullscreen player view controller
window. This happens because UIKit may actually maintain multiple UITextEffectsWindows at
different window levels (only one of which is used to host text selection UI for Live Text in
fullscreen videos). For instance, if the user focuses the unified field in Safari and selects or
edits text in the unified field, a UITextEffectsWindow with a window level of 1.0 will be
instantiated; if this happens prior to triggering Live Text for a fullscreen video, this text
effects window may be reused for Live Text.

In cases where this .windowLevel = 1.0 text effects window is used, the window used for the player
view controller (which currently has a level of UITextEffectsBeneathStatusBarWindowLevel - 1) will
be above the rest of the Live Text selection UI, which causes this bug to occur.

Mitigate this by explicitly grabbing the UITextEffectsWindow that corresponds to the player view's
window using +sharedTextEffectsWindowForWindowScene:, and setting the player view window's level
to a value that's right below this text effects window.

• Source/WebCore/PAL/pal/ios/UIKitSoftLink.h:
• Source/WebCore/PAL/pal/ios/UIKitSoftLink.mm:
• Source/WebCore/PAL/pal/spi/ios/UIKitSPI.h:
• Source/WebCore/platform/ios/VideoFullscreenInterfaceAVKit.mm:

(VideoFullscreenInterfaceAVKit::doSetup):

Canonical link: ​https://commits.webkit.org/251633@main

10:26 PM Changeset in webkit [295627] by Claudio Saavedra

• 4 edits in trunk/Source

Assorted build-time warnings fixes
​https://bugs.webkit.org/show_bug.cgi?id=241681

Reviewed by Mark Lam.

• Source/JavaScriptCore/bytecode/Repatch.cpp:

(JSC::linkPolymorphicCall): Use declared codeBlock variable.

• Source/JavaScriptCore/runtime/TemporalPlainTime.cpp:

(JSC::roundTime): Fix function reaching end without return
value.

• Source/WebCore/css/parser/CSSPropertyParserHelpers.cpp:

(WebCore::CSSPropertyParserHelpers::computeMinimumValue): Ditto.

• Source/WebCore/layout/formattingContexts/flex/FlexLayout.cpp:

(WebCore::Layout::FlexLayout::computeLogicalHeightForFlexItems):
Remove leftover variable.

Canonical link: ​https://commits.webkit.org/251632@main

9:53 PM Changeset in webkit [295626] by Kate Cheney

• 12 edits in trunk

[Trackpad with iPad] Right-clicking on text sometimes doesn't show the correct list of actions
​https://bugs.webkit.org/show_bug.cgi?id=241645
rdar://91792562

Reviewed by Devin Rousso.

This patch fixes a race condition in Reveal code where we may not have updated
the selection in UITextInteractionAssistant before calling the
completion handler for the Reveal menu. This results in an empty menu
appearing often during a right-click.

To fix this, this patch converts sendEditorStateUpdate to an async call and
makes sure that call has completed before returning the context menu.

Since this is a race condition, it was a bit tricky to test. This patch
adds a function _simulateSelectionStart which sets up the WKContentView
as if a selection has begun, but it doesn't actually contact the
UITextInteractionAssistant (to avoid flaky false positives). Then the test calls
the context menu code and makes sure that the UITextInteractionAssistant has
been contacted by the time it completes.

• Source/WebKit/UIProcess/API/Cocoa/WKWebViewPrivateForTesting.h:
• Source/WebKit/UIProcess/API/Cocoa/WKWebViewTesting.mm:

(-[WKWebView _simulateSelectionStart]):

• Source/WebKit/UIProcess/WebPageProxy.cpp:

(WebKit::WebPageProxy::editorStateChanged):

• Source/WebKit/UIProcess/WebPageProxy.h:
• Source/WebKit/UIProcess/WebPageProxy.messages.in:
• Source/WebKit/UIProcess/ios/WKContentViewInteraction.h:
• Source/WebKit/UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView _simulateSelectionStart]):

• Source/WebKit/WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::deleteSurrounding):
(WebKit::WebPage::didApplyStyle):
(WebKit::WebPage::didChangeContents):
(WebKit::WebPage::didUpdateComposition):
(WebKit::WebPage::didEndUserTriggeredSelectionChanges):
(WebKit::WebPage::discardedComposition):
(WebKit::WebPage::canceledComposition):
(WebKit::WebPage::sendEditorStateUpdate):
(WebKit::WebPage::flushPendingEditorStateUpdate):

• Source/WebKit/WebProcess/WebPage/WebPage.h:
• Source/WebKit/WebProcess/WebPage/ios/WebPageIOS.mm:

(WebKit::WebPage::prepareSelectionForContextMenuWithLocationInView):
(WebKit::WebPage::requestPositionInformation):

• Tools/TestWebKitAPI/Tests/WebKitCocoa/iOSMouseSupport.mm:

(handleUpdatedSelection):
(TEST):

• Tools/TestWebKitAPI/ios/UIKitSPI.h:

Canonical link: ​https://commits.webkit.org/251631@main

7:47 PM Changeset in webkit [295625] by commit-queue@webkit.org

• 12 edits
  2 adds in trunk

Fix styling of th elements when explicitly specifiying text-align:inherit
​https://bugs.webkit.org/show_bug.cgi?id=138577

Patch by Ryan Reno <​rreno@apple.com> on 2022-06-16
Reviewed by Tim Nguyen.

<th> elements were being incorrectly centered when specifying
text-align: inherit. This fixes that bug by adding a new internal CSS
value for use in the UA stylesheet. This also removes a non-inherited
flag that was meant to be used for detecting this special case but ultimately didn't
work due to conflicts with the all property.

• Source/WebCore/css/CSSValueKeywords.in:
• Source/WebCore/css/html.css:

(th):

• Source/WebCore/css/parser/CSSParserFastPaths.cpp:

(WebCore::CSSParserFastPaths::isValidKeywordPropertyAndValue):

• Source/WebCore/css/parser/CSSParserIdioms.cpp:

(WebCore::isValueAllowedInMode):

• Source/WebCore/rendering/style/RenderStyle.cpp:

(WebCore::RenderStyle::RenderStyle):

• Source/WebCore/rendering/style/RenderStyle.h:

(WebCore::RenderStyle::NonInheritedFlags::operator== const):
(WebCore::RenderStyle::hasExplicitlySetTextAlign const): Deleted.
(WebCore::RenderStyle::setHasExplicitlySetTextAlign): Deleted.

• Source/WebCore/style/StyleAdjuster.cpp:

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

• Source/WebCore/style/StyleBuilderConverter.h:

(WebCore::Style::BuilderConverter::convertTextAlign):

• Source/WebCore/css/CSSProperties.json:
• Source/WebCore/style/StyleBuilderCustom.h:

(WebCore::Style::BuilderCustom::applyInitialTextAlign): Deleted.
(WebCore::Style::BuilderCustom::applyValueTextAlign): Deleted.

• LayoutTests/fast/css/internal-th-center-ua-only-expected.txt: Added.
• LayoutTests/fast/css/internal-th-center-ua-only.html: Added.
• LayoutTests/fast/table/center-th-when-parent-has-initial-text-align-expected.html:
• LayoutTests/fast/table/center-th-when-parent-has-initial-text-align.html:

Canonical link: ​https://commits.webkit.org/251630@main

7:20 PM Changeset in webkit [295624] by basuke.suzuki@sony.com

• 5 edits in trunk/Source/JavaScriptCore

The extraMemorySize() get wrong when transferring ArrayBuffer from Worker VM
​https://bugs.webkit.org/show_bug.cgi?id=241559

Reviewed by Yusuke Suzuki.

When ArrayBuffer is passed in the transfer option of postMessage(), the size cached in
heap.m_arrayBuffers get incorrect and that makes extraMemorySize() bigger than actual
managed size.

This patch added the code to reduce size from GCIncomingRefCountedSet.m_bytes when
ArrayBuffer is actually transferring from VM.

Also for verification, added a simple check code in GCIncomingRefCountedSet.addReference
with constexpr flag.

• Source/JavaScriptCore/heap/GCIncomingRefCountedSet.h:
• Source/JavaScriptCore/heap/GCIncomingRefCountedSetInlines.h:

(JSC::GCIncomingRefCountedSet<T>::sweep):
(JSC::GCIncomingRefCountedSet<T>::reduceSize):

• Source/JavaScriptCore/heap/Heap.cpp:

(JSC::Heap::reduceArrayBufferSize):

• Source/JavaScriptCore/heap/Heap.h:
• Source/JavaScriptCore/runtime/ArrayBuffer.cpp:

(JSC::ArrayBuffer::transferTo):

Canonical link: ​https://commits.webkit.org/251629@main

7:17 PM Changeset in webkit [295623] by Devin Rousso

• 13 edits
  2 adds in trunk

Web Inspector: Elements: rework CSS pseudo class toggles
​https://bugs.webkit.org/show_bug.cgi?id=241655

Reviewed by Patrick Angle.

Move the forced pseudo class checkboxes to the bottom of the sidebar next to the class list
checkboxes (though only one of them can be enabled at a time) since more space is needed to support
new (and future) pseudo classes and using space at the bottom of the sidebar is better than at the
top (since there's already precedent with the class list checkboxes).

Test: inspector/css/forcePseudoState.html

• Source/WebInspectorUI/UserInterface/Views/GeneralStyleDetailsSidebarPanel.js:

(WI.GeneralStyleDetailsSidebarPanel):
(WI.GeneralStyleDetailsSidebarPanel.prototype.get minimumWidth):
(WI.GeneralStyleDetailsSidebarPanel.prototype.attached):
(WI.GeneralStyleDetailsSidebarPanel.prototype.layout):
(WI.GeneralStyleDetailsSidebarPanel.prototype.addEventListeners):
(WI.GeneralStyleDetailsSidebarPanel.prototype.removeEventListeners):
(WI.GeneralStyleDetailsSidebarPanel.prototype.initialLayout):
(WI.GeneralStyleDetailsSidebarPanel.prototype.sizeDidChange):
(WI.GeneralStyleDetailsSidebarPanel.prototype._updateClassListContainer): Added.
(WI.GeneralStyleDetailsSidebarPanel.prototype._updateForcedPseudoClassContainer): Added.
(WI.GeneralStyleDetailsSidebarPanel.prototype._handleNodeChanged):
(WI.GeneralStyleDetailsSidebarPanel.prototype._forcedPseudoClassCheckboxChanged): Added.
(WI.GeneralStyleDetailsSidebarPanel.prototype._updatePseudoClasasCheckboxes): Added.
(WI.GeneralStyleDetailsSidebarPanel.prototype._classListToggleButtonClicked): Renamed from _classToggleButtonClicked.
(WI.GeneralStyleDetailsSidebarPanel.prototype._forcedPseudoClassToggleButtonClicked): Added.
(WI.GeneralStyleDetailsSidebarPanel.prototype.styleDetailsPanelFocusLastPseudoClassCheckbox): Deleted.
(WI.GeneralStyleDetailsSidebarPanel.prototype.get _initialScrollOffset): Deleted.
(WI.GeneralStyleDetailsSidebarPanel.prototype._updateNoForcedPseudoClassesScrollOffset): Deleted.
(WI.GeneralStyleDetailsSidebarPanel.prototype._handleForcedPseudoClassCheckboxKeydown): Deleted.

• Source/WebInspectorUI/UserInterface/Views/GeneralStyleDetailsSidebarPanel.css:

(.sidebar > .panel.details.css-style > .content ~ :is(.options-container, .class-list-container, .forced-pseudo-class-container)): ADded.
(.sidebar > .panel.details.css-style > .content ~ :is(.options-container, .class-list-container)):
(.sidebar > .panel.details.css-style > .content:not(.supports-new-rule, .has-filter-bar) ~ :is(.options-container, .class-list-container, .forced-pseudo-class-container)): Renamed from .sidebar > .panel.details.css-style > .content:not(.supports-new-rule, .has-filter-bar) ~ :is(.options-container, .class-list-container).
(.sidebar > .panel.details.css-style > .content ~ .options-container > .toggle): Renamed from .sidebar > .panel.details.css-style > .content ~ .options-container > .toggle-class-toggle.
(.sidebar > .panel.details.css-style > .content ~ .options-container > .toggle:focus): Renamed from .sidebar > .panel.details.css-style > .content ~ .options-container > .toggle-class-toggle:focus.
(.sidebar > .panel.details.css-style > .content ~ .options-container > .toggle::before): Renamed from .sidebar > .panel.details.css-style > .content ~ .options-container > .toggle-class-toggle::before.
(.sidebar > .panel.details.css-style > .content ~ .options-container > .toggle:is(.selected, :hover)): Renamed from .sidebar > .panel.details.css-style > .content ~ .options-container > .toggle-class-toggle:is(.selected, :hover).
(.sidebar > .panel.details.css-style > .content ~ .options-container > .toggle:is(.selected, :hover)::before): Renamed from .sidebar > .panel.details.css-style > .content ~ .options-container > .toggle-class-toggle:is(.selected, :hover)::before.
(.sidebar > .panel.details.css-style > .content ~ .options-container > .toggle:not(.selected):hover::before): Renamed from .sidebar > .panel.details.css-style > .content ~ .options-container > .toggle-class-toggle:not(.selected):hover::before.
(.sidebar > .panel.details.css-style > .content ~ .options-container > .toggle.selected:active::before): Renamed from .sidebar > .panel.details.css-style > .content ~ .options-container > .toggle-class-toggle.selected:active::before.
(.sidebar > .panel.details.css-style > .content:not(.supports-new-rule) ~ .options-container > .new-rule, .sidebar > .panel.details.css-style > .content:not(.supports-toggle-class-list) ~ .options-container > .toggle.class-list, .sidebar > .panel.details.css-style > .content:not(.supports-toggle-forced-pseudo-class) ~ .options-container > .toggle.forced-pseudo-class, .sidebar > .panel.details.css-style > .content:not(.has-filter-bar) ~ .options-container > .filter-bar, .sidebar > .panel.details.css-style > .content:not(.supports-new-rule):not(.supports-toggle-class-list):not(.supports-toggle-forced-pseudo-class):not(.has-filter-bar) ~ .options-container): Renamed from .sidebar > .panel.details.css-style > .content:not(.supports-new-rule) ~ .options-container > .new-rule, .sidebar > .panel.details.css-style > .content:not(.supports-toggle-css-class) ~ .options-container > .toggle-class-toggle, .sidebar > .panel.details.css-style > .content:not(.has-filter-bar) ~ .options-container > .filter-bar, .sidebar > .panel.details.css-style > .content:not(.supports-new-rule):not(.supports-toggle-class):not(.has-filter-bar) ~ .options-container.
(.sidebar > .panel.details.css-style > .content ~ :is(.class-list-container, .forced-pseudo-class-container)): Added.
(.sidebar > .panel.details.css-style > .content ~ :is(.class-list-container, .forced-pseudo-class-container)[hidden]): Added.
(.sidebar > .panel.details.css-style > .content ~ .class-list-container):
(.sidebar > .panel.details.css-style > .content ~ .forced-pseudo-class-container): Added.
(.sidebar > .panel.details.css-style > .content ~ .forced-pseudo-class-container > label): Added.

• Source/WebInspectorUI/UserInterface/Views/SpreadsheetRulesStyleDetailsPanel.js:

(WI.SpreadsheetRulesStyleDetailsPanel.prototype.get supportsToggleCSSClassList): Added.
(WI.SpreadsheetRulesStyleDetailsPanel.prototype.get supportsToggleCSSForcedPseudoClass): Added.
(WI.SpreadsheetRulesStyleDetailsPanel.prototype.get initialToggleCSSForcedPseudoClassState): Added.
(WI.SpreadsheetRulesStyleDetailsPanel.prototype.spreadsheetCSSStyleDeclarationSectionStartEditingAdjacentRule):
Remove styleDetailsPanelFocusLastPseudoClassCheckbox as the forced pseudo class checkboxes are no
longer at the top, and therefore should not be focused when shift-tabbing from the first property.

• Source/WebInspectorUI/UserInterface/Views/StyleDetailsPanel.js:

(WI.StyleDetailsPanel.prototype.get supportsToggleCSSClassList): Renamed from get supportsToggleCSSClass.
(WI.StyleDetailsPanel.prototype.get supportsToggleCSSForcedPseudoClass):
(WI.StyleDetailsPanel.prototype.get _initialScrollOffset): Deleted.
No need to adjust the initial scrollTop for the forced pseudo class checkboxes since they are no
longer at the top.

• Source/WebInspectorUI/UserInterface/Views/ComputedStyleDetailsPanel.js:

(WI.ComputedStyleDetailsPanel.prototype.get supportsToggleCSSClassList): Added.
(WI.ComputedStyleDetailsPanel.prototype.get supportsToggleCSSForcedPseudoClass): Added.
(WI.ComputedStyleDetailsPanel.prototype.get initialToggleCSSForcedPseudoClassState): Added.

• Source/WebInspectorUI/UserInterface/Views/FontDetailsPanel.js:

(WI.FontDetailsPanel.prototype.get supportsToggleCSSClass): Deleted.
Allow WI.StyleDetailsPanel subclasses to control whether they support forcing pseudo classes (and
if so whether the container should initially be shown).

• Source/JavaScriptCore/inspector/protocol/CSS.json:
• Source/WebCore/inspector/agents/InspectorCSSAgent.h:
• Source/WebCore/inspector/agents/InspectorCSSAgent.cpp:

(WebCore::computePseudoClassMask): Deleted.
(WebCore::InspectorCSSAgent::forcePseudoState):

• Source/WebInspectorUI/UserInterface/Controllers/CSSManager.js:

(WI.CSSManager.displayNameForForceablePseudoClass): Added.
(WI.CSSManager.prototype.canForcePseudoClass): Renamed from canForcePseudoClasses.
Add enum values for :focus-visible and :focus-within.
Drive-by: Make CSS.ForceablePseudoClass enum instead of having it be an inline/anonymous enum for

only CSS.forcePseudoState.

• Source/WebInspectorUI/UserInterface/Views/ContextMenuUtilities.js:

(WI.appendContextMenuItemsForDOMNode):

• Source/WebInspectorUI/UserInterface/Views/SpreadsheetCSSStyleDeclarationSection.js:

(WI.SpreadsheetCSSStyleDeclarationSection.prototype._populateIconElementContextMenu):
Handle WI.CSSManager.ForceablePseudoClass now being an object instead of an array.

• Source/WebInspectorUI/Localizations/en.lproj/localizedStrings.js:

• LayoutTests/inspector/css/forcePseudoState.html: Added.
• LayoutTests/inspector/css/forcePseudoState-expected.txt: Added.

Canonical link: ​https://commits.webkit.org/251628@main

6:43 PM Changeset in webkit [295622] by ysuzuki@apple.com

• 1 edit in trunk/.github/CODEOWNERS

Unreviewed, add bmalloc related information to .github/CODEOWNERS

• .github/CODEOWNERS:

Canonical link: ​https://commits.webkit.org/251627@main

6:38 PM Changeset in webkit [295621] by Cameron McCormack

• 64 edits
  1 copy
  15 adds in trunk

Add a new DrawDecomposedGlyphs display list item to avoid repeatedly sending glyphs when using the GlyphDisplayListCache
​https://bugs.webkit.org/show_bug.cgi?id=240497
<rdar://93387615>

Reviewed by Simon Fraser.

The GlyphDisplayListCache is used to record a display list for
frequently painting text content. With GPU Process DOM rendering, there
is significant overhead in sending the contents of these display lists
over IPC. The contents of these display lists don't change if the text
content in the document doesn't change, so we could greatly reduce the
overhead by treating the data inside a display list item for glyph
drawing as a remote resource.

This commit adds:

• a new display list item, DrawDecomposedGlyphs, to represent drawing a glyph list resource
• a new class, DecomposedGlyphs, which is the resource type
• a new struct, PositionedGlyphs, to provide a common place for the glyph drawing fields (the vector of glyph IDs, the anchor position, etc.) to live, so that we don't have duplication between DisplayList::DrawGlyphs and DecomposedGlyphs

So that a DrawDecomposedGlyphs command can be replayed from a
GlyphDisplayListCache's in-memory display list and recorded to a
RemoteDisplayListRecorder, the GraphicsContext API gains a new
drawDecomposedGlyphs function.

A new argument to the DisplayList::RecordImpl constructor (and the
DrawGlyphsRecorder) is added to represent how to record drawText
commands:

• DrawGlyphsMode::Normal, which records each GraphicsContext::drawText call with a single DrawText command
• DrawGlyphsMode::DeconstructToDrawGlyphsCommands, which ensures different text layers get deconstructed into separate DrawText commands
• DrawGlyphsMode::DeconstructToDrawDecomposedGlyphsCommands, which ensures different text layers get desconstructed into separate DrawDecomposedGlyphs commands

FontCascade::displayListForTextRun is updated to use that last value.

Additionally, GlyphDisplayListCache is extended to cache display lists
keyed off TextRun/FontCascade/etc. values. This allows sharing of the same
cached display list between different elements on the page that have the same
text content.

This sharing would not be valid if the two elements have different
values for the color property, and the text contains COLRv0 glyphs that
alternate painting of specific colors and the color fill color, since
the recording would incorrectly record a setFillBrush command
corresponding to the first element's fill color. Rather than extend the
glyph recorder to parameterize the current fill (and stroke) colors, we
detect when outlines are drawn with colors other than the context's
initial colors, and prevent sharing. This is done by checking whether
the recorded display list contains items that aren't known to be safe
for sharing.

Similarly, if the sharing would not be valid if the contains bitmap
images (like those from emoji fonts) or SVG glyphs, both of which are
captured as DrawNativeImage commands, if the text is drawn at different
scales. This is because the size of the images is dependent on the
scale. We detect and prevent reuse across different text runs if the
scale is different, by checking the recorded display list for
DrawNativeImage commands and by storing the context scale on the
GlyphDisplayListCache::Entry.

• LayoutTests/fast/text/glyph-display-lists/glyph-display-list-color-expected.txt:
• LayoutTests/fast/text/glyph-display-lists/glyph-display-list-colr-unshared-expected.txt: Added.
• LayoutTests/fast/text/glyph-display-lists/glyph-display-list-colr-unshared.html: Added.
• LayoutTests/fast/text/glyph-display-lists/glyph-display-list-scaled-unshared-expected.txt: Added.
• LayoutTests/fast/text/glyph-display-lists/glyph-display-list-scaled-unshared.html: Added.
• LayoutTests/fast/text/glyph-display-lists/glyph-display-list-shadow-unshared-expected.txt: Added.
• LayoutTests/fast/text/glyph-display-lists/glyph-display-list-shadow-unshared.html: Added.
• LayoutTests/fast/text/glyph-display-lists/glyph-display-list-shared-expected.txt: Added.
• LayoutTests/fast/text/glyph-display-lists/glyph-display-list-shared.html: Added.
• LayoutTests/fast/text/glyph-display-lists/glyph-display-list-svg-unshared-expected.txt: Added.
• LayoutTests/fast/text/glyph-display-lists/glyph-display-list-svg-unshared.html: Added.
• Source/WTF/wtf/PlatformHave.h:
• Source/WebCore/Headers.cmake:
• Source/WebCore/Sources.txt:
• Source/WebCore/WebCore.xcodeproj/project.pbxproj:
• Source/WebCore/page/MemoryRelease.cpp:

(WebCore::releaseNoncriticalMemory):

• Source/WebCore/platform/graphics/BifurcatedGraphicsContext.cpp:

(WebCore::BifurcatedGraphicsContext::drawDecomposedGlyphs):

• Source/WebCore/platform/graphics/BifurcatedGraphicsContext.h:
• Source/WebCore/platform/graphics/DecomposedGlyphs.cpp: Added.

(WebCore::DecomposedGlyphs::create):
(WebCore::DecomposedGlyphs::DecomposedGlyphs):
(WebCore::m_renderingResourceIdentifier):

• Source/WebCore/platform/graphics/DecomposedGlyphs.h: Added.

(WebCore::DecomposedGlyphs::positionedGlyphs const):
(WebCore::DecomposedGlyphs::bounds const):
(WebCore::DecomposedGlyphs::addObserver):
(WebCore::DecomposedGlyphs::removeObserver):
(WebCore::DecomposedGlyphs::renderingResourceIdentifier const):

• Source/WebCore/platform/graphics/FontCascade.cpp:

(WebCore::FontCascade::displayListForTextRun const):

• Source/WebCore/platform/graphics/GraphicsContext.cpp:

(WebCore::GraphicsContext::drawDecomposedGlyphs):

• Source/WebCore/platform/graphics/GraphicsContext.h:

(WebCore::GraphicsContext::drawGlyphsAndCacheResources):
(WebCore::GraphicsContext::drawGlyphsAndCacheFont): Deleted.

• Source/WebCore/platform/graphics/NullGraphicsContext.h:
• Source/WebCore/platform/graphics/PositionedGlyphs.cpp: Copied from Source/WebCore/platform/graphics/win/DrawGlyphsRecorderWin.cpp.

(WebCore::PositionedGlyphs::computeBounds const):

• Source/WebCore/platform/graphics/PositionedGlyphs.h: Added.

(WebCore::PositionedGlyphs::PositionedGlyphs):
(WebCore::PositionedGlyphs::encode const):
(WebCore::PositionedGlyphs::decode):

• Source/WebCore/platform/graphics/TextRun.cpp:

(WebCore::operator<<):

• Source/WebCore/platform/graphics/TextRun.h:

(WebCore::TextRun::TextRun):
(WebCore::TextRun::isHashTableEmptyValue const):
(WebCore::TextRun::isHashTableDeletedValue const):
(WebCore::TextRun::isolatedCopy const):

• Source/WebCore/platform/graphics/TextRunHash.h: Added.

(WebCore::add):
(WebCore::TextRun::operator== const):
(WebCore::TextRunHash::hash):
(WebCore::TextRunHash::equal):
(WTF::HashTraits<WebCore::TextRun>::isDeletedValue):
(WTF::HashTraits<WebCore::TextRun>::isEmptyValue):
(WTF::HashTraits<WebCore::TextRun>::constructDeletedValue):
(WTF::HashTraits<WebCore::TextRun>::emptyValue):

• Source/WebCore/platform/graphics/cairo/GraphicsContextCairo.cpp:

(WebCore::GraphicsContextCairo::drawDecomposedGlyphs):

• Source/WebCore/platform/graphics/cairo/GraphicsContextCairo.h:
• Source/WebCore/platform/graphics/coretext/DrawGlyphsRecorderCoreText.cpp:

(WebCore::DrawGlyphsRecorder::createInternalContext):
(WebCore::DrawGlyphsRecorder::updateCTM):
(WebCore::DrawGlyphsRecorder::recordDrawGlyphs):

• Source/WebCore/platform/graphics/displaylists/DisplayList.cpp:

(WebCore::DisplayList::DisplayList::description const):
(WebCore::DisplayList::DisplayList::append):

• Source/WebCore/platform/graphics/displaylists/DisplayList.h:

(WebCore::DisplayList::DisplayList::cacheDecomposedGlyphs):

• Source/WebCore/platform/graphics/displaylists/DisplayListItemBuffer.cpp:

(WebCore::DisplayList::ItemHandle::apply):
(WebCore::DisplayList::ItemHandle::destroy):
(WebCore::DisplayList::ItemHandle::safeCopy const):

• Source/WebCore/platform/graphics/displaylists/DisplayListItemType.cpp:

(WebCore::DisplayList::sizeOfItemInBytes):
(WebCore::DisplayList::isDrawingItem):
(WebCore::DisplayList::isInlineItem):

• Source/WebCore/platform/graphics/displaylists/DisplayListItemType.h:
• Source/WebCore/platform/graphics/displaylists/DisplayListItems.cpp:

(WebCore::DisplayList::DrawGlyphs::DrawGlyphs):
(WebCore::DisplayList::m_bounds):
(WebCore::DisplayList::DrawGlyphs::apply const):
(WebCore::DisplayList::DrawDecomposedGlyphs::apply const):
(WebCore::DisplayList::operator<<):
(WebCore::DisplayList::dumpItem):
(WebCore::DisplayList::dumpItemHandle):
(WebCore::DisplayList::DrawGlyphs::computeBounds): Deleted.

• Source/WebCore/platform/graphics/displaylists/DisplayListItems.h:

(WebCore::DisplayList::DrawGlyphs::localAnchor const):
(WebCore::DisplayList::DrawGlyphs::anchorPoint const):
(WebCore::DisplayList::DrawGlyphs::glyphs const):
(WebCore::DisplayList::DrawGlyphs::encode const):
(WebCore::DisplayList::DrawGlyphs::decode):
(WebCore::DisplayList::DrawDecomposedGlyphs::DrawDecomposedGlyphs):
(WebCore::DisplayList::DrawDecomposedGlyphs::fontIdentifier const):
(WebCore::DisplayList::DrawDecomposedGlyphs::decomposedGlyphsIdentifier const):
(WebCore::DisplayList::DrawDecomposedGlyphs::globalBounds const):
(WebCore::DisplayList::DrawDecomposedGlyphs::localBounds const):

• Source/WebCore/platform/graphics/displaylists/DisplayListRecorder.cpp:

(WebCore::DisplayList::Recorder::Recorder):
(WebCore::DisplayList::Recorder::shouldDeconstructDrawGlyphs const):
(WebCore::DisplayList::Recorder::drawGlyphs):
(WebCore::DisplayList::Recorder::drawDecomposedGlyphs):
(WebCore::DisplayList::Recorder::drawGlyphsAndCacheResources):
(WebCore::DisplayList::Recorder::drawGlyphsAndCacheFont): Deleted.

• Source/WebCore/platform/graphics/displaylists/DisplayListRecorder.h:
• Source/WebCore/platform/graphics/displaylists/DisplayListRecorderImpl.cpp:

(WebCore::DisplayList::RecorderImpl::RecorderImpl):
(WebCore::DisplayList::RecorderImpl::recordDrawDecomposedGlyphs):
(WebCore::DisplayList::RecorderImpl::recordResourceUse):

• Source/WebCore/platform/graphics/displaylists/DisplayListRecorderImpl.h:
• Source/WebCore/platform/graphics/displaylists/DisplayListReplayer.cpp:

(WebCore::DisplayList::applyDrawDecomposedGlyphs):
(WebCore::DisplayList::Replayer::applyItem):
(WebCore::DisplayList::Replayer::replay):

• Source/WebCore/platform/graphics/displaylists/DisplayListReplayer.h:
• Source/WebCore/platform/graphics/displaylists/DisplayListResourceHeap.h:

(WebCore::DisplayList::LocalResourceHeap::add):

• Source/WebCore/platform/graphics/harfbuzz/DrawGlyphsRecorderHarfBuzz.cpp:

(WebCore::DrawGlyphsRecorder::drawGlyphs):

• Source/WebCore/platform/graphics/nicosia/cairo/NicosiaCairoOperationRecorder.cpp:

(Nicosia::CairoOperationRecorder::drawDecomposedGlyphs):

• Source/WebCore/platform/graphics/nicosia/cairo/NicosiaCairoOperationRecorder.h:
• Source/WebCore/platform/graphics/win/DrawGlyphsRecorderWin.cpp:

(WebCore::DrawGlyphsRecorder::drawGlyphs):

• Source/WebCore/platform/text/TextDirection.h:

(WebCore::operator<<):

• Source/WebCore/platform/text/TextFlags.cpp:

(WebCore::operator<<):

• Source/WebCore/platform/text/TextFlags.h:

(WebCore::ExpansionBehavior::operator== const):

• Source/WebCore/rendering/GlyphDisplayListCache.cpp: Added.

(WebCore::canShareDisplayListWithItem):
(WebCore::add):
(WebCore::GlyphDisplayListCacheKeyTranslator::hash):
(WebCore::GlyphDisplayListCacheKeyTranslator::equal):
(WebCore::GlyphDisplayListCache::singleton):
(WebCore::GlyphDisplayListCache::clear):
(WebCore::GlyphDisplayListCache::size const):
(WebCore::GlyphDisplayListCache::sizeInBytes const):
(WebCore::GlyphDisplayListCache::get):
(WebCore::GlyphDisplayListCache::getIfExists):
(WebCore::GlyphDisplayListCache::remove):
(WebCore::GlyphDisplayListCache::canShareDisplayList):
(WebCore::GlyphDisplayListCacheEntry::~GlyphDisplayListCacheEntry):

• Source/WebCore/rendering/GlyphDisplayListCache.h:

(WebCore::GlyphDisplayListCacheEntry::create):
(WebCore::GlyphDisplayListCacheEntry::operator== const):
(WebCore::GlyphDisplayListCacheEntry::displayList):
(WebCore::GlyphDisplayListCacheEntry::GlyphDisplayListCacheEntry):
(WebCore::add):
(WebCore::GlyphDisplayListCacheEntryHash::hash):
(WebCore::GlyphDisplayListCacheEntryHash::equal):
(WebCore::GlyphDisplayListCache::get):
(WebCore::GlyphDisplayListCache::getIfExists):
(WebCore::GlyphDisplayListCache::remove):
(WebCore::GlyphDisplayListCache::singleton): Deleted.
(WebCore::GlyphDisplayListCache::clear): Deleted.
(WebCore::GlyphDisplayListCache::size const): Deleted.
(WebCore::GlyphDisplayListCache::sizeInBytes const): Deleted.

• Source/WebCore/rendering/RenderLayerCompositor.cpp:
• Source/WebCore/rendering/TextPainter.cpp:

(WebCore::TextPainter::clearGlyphDisplayLists): Deleted.

• Source/WebCore/rendering/TextPainter.h:

(WebCore::TextPainter::setGlyphDisplayListIfNeeded):
(WebCore::TextPainter::removeGlyphDisplayList):
(WebCore::TextPainter::glyphDisplayListIfExists):

• Source/WebCore/testing/Internals.cpp:

(WebCore::toDisplayListFlags):
(WebCore::Internals::displayListForElement):
(WebCore::Internals::replayDisplayListForElement):
(WebCore::Internals::cachedGlyphDisplayListsForTextNode):

• Source/WebCore/testing/Internals.h:
• Source/WebCore/testing/Internals.idl:
• Source/WebKit/GPUProcess/graphics/QualifiedResourceHeap.h:

(WebKit::QualifiedResourceHeap::add):
(WebKit::QualifiedResourceHeap::getDecomposedGlyphs const):
(WebKit::QualifiedResourceHeap::removeDecomposedGlyphs):
(WebKit::QualifiedResourceHeap::checkInvariants const):

• Source/WebKit/GPUProcess/graphics/RemoteDisplayListRecorder.cpp:

(WebKit::RemoteDisplayListRecorder::drawDecomposedGlyphs):
(WebKit::RemoteDisplayListRecorder::drawDecomposedGlyphsWithQualifiedIdentifiers):

• Source/WebKit/GPUProcess/graphics/RemoteDisplayListRecorder.h:
• Source/WebKit/GPUProcess/graphics/RemoteDisplayListRecorder.messages.in:
• Source/WebKit/GPUProcess/graphics/RemoteRenderingBackend.cpp:

(WebKit::RemoteRenderingBackend::cacheFontWithQualifiedIdentifier):
(WebKit::RemoteRenderingBackend::cacheDecomposedGlyphs):
(WebKit::RemoteRenderingBackend::cacheDecomposedGlyphsWithQualifiedIdentifier):

• Source/WebKit/GPUProcess/graphics/RemoteRenderingBackend.h:
• Source/WebKit/GPUProcess/graphics/RemoteRenderingBackend.messages.in:
• Source/WebKit/GPUProcess/graphics/RemoteResourceCache.cpp:

(WebKit::RemoteResourceCache::cacheDecomposedGlyphs):
(WebKit::RemoteResourceCache::cachedDecomposedGlyphs const):
(WebKit::RemoteResourceCache::releaseRemoteResource):

• Source/WebKit/GPUProcess/graphics/RemoteResourceCache.h:
• Source/WebKit/Scripts/webkit/messages.py:
• Source/WebKit/Shared/WebCoreArgumentCoders.cpp:

(IPC::ArgumentCoder<DecomposedGlyphs>::encode):
(IPC::ArgumentCoder<DecomposedGlyphs>::decode):

• Source/WebKit/Shared/WebCoreArgumentCoders.h:
• Source/WebKit/WebProcess/GPU/graphics/RemoteDisplayListRecorderProxy.cpp:

(WebKit::RemoteDisplayListRecorderProxy::RemoteDisplayListRecorderProxy):
(WebKit::RemoteDisplayListRecorderProxy::recordDrawDecomposedGlyphs):
(WebKit::RemoteDisplayListRecorderProxy::recordResourceUse):

• Source/WebKit/WebProcess/GPU/graphics/RemoteDisplayListRecorderProxy.h:
• Source/WebKit/WebProcess/GPU/graphics/RemoteRenderingBackendProxy.cpp:

(WebKit::RemoteRenderingBackendProxy::cacheDecomposedGlyphs):

• Source/WebKit/WebProcess/GPU/graphics/RemoteRenderingBackendProxy.h:
• Source/WebKit/WebProcess/GPU/graphics/RemoteResourceCacheProxy.cpp:

(WebKit::RemoteResourceCacheProxy::~RemoteResourceCacheProxy):
(WebKit::RemoteResourceCacheProxy::recordDecomposedGlyphsUse):
(WebKit::RemoteResourceCacheProxy::releaseDecomposedGlyphs):
(WebKit::RemoteResourceCacheProxy::clearDecomposedGlyphsMap):
(WebKit::RemoteResourceCacheProxy::remoteResourceCacheWasDestroyed):

• Source/WebKit/WebProcess/GPU/graphics/RemoteResourceCacheProxy.h:

Canonical link: ​https://commits.webkit.org/251626@main

6:11 PM Changeset in webkit [295620] by dino@apple.com

• 3 edits in trunk/Source

Clean up code that protects volume from being set on iOS
​https://bugs.webkit.org/show_bug.cgi?id=241657
<rdar://problem/95252820>

Reviewed by Eric Carlson.

The code to protect HTMLMediaElement.volume from being writable is guarded
by some PLATFORM(IOS_FAMILY) tests. Make this a
HAVE(MEDIA_VOLUME_PER_ELEMENT) test so it can be expanded a bit easier.

• Source/WTF/wtf/PlatformHave.h:
• Source/WebCore/html/HTMLMediaElement.cpp:

(WebCore::HTMLMediaElement::setVolume):
(WebCore::HTMLMediaElement::updateVolume):
(WebCore::HTMLMediaElement::cancelPendingTasks):
(WebCore::HTMLMediaElement::mediaVolumeDidChange):

• Source/WebCore/platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.mm:

(WebCore::MediaPlayerPrivateAVFoundationObjC::setVolume):

Canonical link: ​https://commits.webkit.org/251625@main

5:51 PM Changeset in webkit [295619] by Cameron McCormack

• 2 edits in trunk/Source/WebCore/platform/graphics

Stop updating FontCascade::{m_useBackslashAsYenSymbol,m_enableKerning,m_requiresShaping} in FontCascade::updateFonts
​https://bugs.webkit.org/show_bug.cgi?id=241672

Unreviewed.

(Actually this was reviewed by Myles Maxfield, but the bots are having
trouble identifying the reviewer.)

These fields are computed based on the FontDescription. They're only mutable
because we also update them in FontCascade::updateFonts(), but that function
only updates the FontCascadeFonts pointer, not the FontDescription. So there
should be no need to update them at that time.

• Source/WebCore/platform/graphics/FontCascade.cpp:

(WebCore::FontCascade::updateFonts const):

• Source/WebCore/platform/graphics/FontCascade.h:

Canonical link: ​https://commits.webkit.org/251624@main

5:21 PM Changeset in webkit [295618] by Tyler Wilcock

• 1 edit in trunk/Source/WebCore/accessibility/AccessibilityObject.cpp

AX: AccessibilityObject::insertChild does not check the validity of the insertionIndex while processing grandchildren
​https://bugs.webkit.org/show_bug.cgi?id=241650

Reviewed by Chris Fleizach.

When AccessibilityObject::insertChild is asked to insert a child that's
ignored, we instead add that object's children. However, both
accessibilityIsIgnored and children can cause layout, and said
layout could cause AccessibilityObject::m_children to be cleared. This
makes the insertionIndex invalid, which causes a crash.

In this patch, right before m_children.insert(), we check to make sure
the index is still valid.

I wasn't able to make a test for this bug. It is difficult to reproduce,
and the circumstances to reproduce are complex.

• Source/WebCore/accessibility/AccessibilityObject.cpp:

(WebCore::AccessibilityObject::insertChild):

Canonical link: ​https://commits.webkit.org/251623@main

4:52 PM Changeset in webkit [295617] by J Pascoe

• 1 edit in trunk/Source/WebKit/UIProcess/WebAuthentication/Cocoa/LocalAuthenticator.mm

[WebAuthn] Stop using decidePolicyForLocalAuthenticator
​https://bugs.webkit.org/show_bug.cgi?id=241614
rdar://95066808

Reviewed by Brent Fulgham.

• Source/WebKit/UIProcess/WebAuthentication/Cocoa/LocalAuthenticator.mm:

(WebKit::LocalAuthenticator::makeCredential):
Previously, in the browser ui delegate based flow, consent would be given
for the platform authenticator after selecting it. In the new flow you
must consent to the platform authenticator before getting here, so
decidePolicyForLocalAuthenticator is no longer needed.

Canonical link: ​https://commits.webkit.org/251622@main

4:37 PM Changeset in webkit [295616] by J Pascoe

• 6 edits in trunk/Source

[WebAuthn] Rename cable transport to hybrid
​https://bugs.webkit.org/show_bug.cgi?id=241691
rdar://problem/95312126

Reviewed by Brent Fulgham.

This transport got renamed in ​https://github.com/fido-alliance/fido-2-specs/issues/1332

Hybrid is the name that should be used in AuthenticatorTransport, we continue to recgnize
the old string.

• Source/WebCore/Modules/webauthn/AuthenticatorTransport.h:
• Source/WebCore/Modules/webauthn/fido/AuthenticatorGetInfoResponse.cpp:

(fido::toString):

• Source/WebCore/Modules/webauthn/fido/DeviceResponseConverter.cpp:

(fido::convertStringToAuthenticatorTransport):

• Source/WebKit/UIProcess/WebAuthentication/Cocoa/LocalAuthenticator.mm:

(WebKit::LocalAuthenticatorInternal::transports):

• Source/WebKit/UIProcess/WebAuthentication/Cocoa/WebAuthenticatorCoordinatorProxy.mm:

(WebKit::toASCDescriptor):

Canonical link: ​https://commits.webkit.org/251621@main

4:23 PM Changeset in webkit [295615] by Aditya Keerthi

• 2 edits in trunk/Source

[iOS] Update the title of the context menu action to save an image
​https://bugs.webkit.org/show_bug.cgi?id=241690
rdar://95117013

Reviewed by Kate Cheney.

Change the title from "Add to Photos" to "Save to Photos" to align with
strings used in the Photos app.

• Source/WebCore/en.lproj/Localizable.strings:
• Source/WebKit/UIProcess/API/Cocoa/_WKElementAction.mm:

(+[_WKElementAction _elementActionWithType:customTitle:assistant:]):

Canonical link: ​https://commits.webkit.org/251620@main

4:08 PM Changeset in webkit [295614] by ysuzuki@apple.com

• 10 edits
  1 add in trunk

[JSC] Always create StructureStubInfo for op_get_by_val
​https://bugs.webkit.org/show_bug.cgi?id=241669
rdar://75146284

Reviewed by Saam Barati and Mark Lam.

DFG OSR exit requires StructureStubInfo for getter / setter calls. However very generic baseline JIT
op_get_by_val does not create StructureStubInfo. It is possible that OSR exit crashes because of this
missing StructureStubInfo. Let's consider the following edge case.

1. Now, Baseline detects that this is very generic op_get_by_val. So we do not create StructureStubInfo.
2. This function is inlined in DFG. And DFG emits IC for this GetByVal.
3. (2)'s DFG function collects information in DFG-level IC. And luckily, in this inlined call path, it was not so generic.
4. Then, due to different OSR exit or something, we recreate DFG code for this function with (2)'s inlining.
5. DFG detects that DFG-level IC has more specialized information. So it can inline getter call in this op_get_by_val.
6. Inside this getter, we perform OSR exit.
7. Looking into Baseline, and we found that there is no StructureStubInfo!

We always create StructureStubInfo. In very generic op_get_by_val case, we create this with tookSlowPath = true.
And we emit empty inline path to record doneLocation. So, OSR exit can jump to this place.

We also clean up StructureStubInfo code.

1. "start" is renamed to startLocation. And we do not record it in DataIC case since it is not necessary.
2. Rename inlineSize to inlineCodeSize.
3. Add some assertions to ensure that this path is not used for DataIC case.
4. We also record opcode value in the crashing RELEASE_ASSERT to get more information if this does not fix the issue.

• Source/JavaScriptCore/bytecode/InlineAccess.cpp:

(JSC::linkCodeInline):
(JSC::InlineAccess::generateArrayLength):
(JSC::InlineAccess::generateStringLength):
(JSC::InlineAccess::rewireStubAsJumpInAccessNotUsingInlineAccess):
(JSC::InlineAccess::rewireStubAsJumpInAccess):
(JSC::InlineAccess::resetStubAsJumpInAccess):

• Source/JavaScriptCore/bytecode/StructureStubInfo.cpp:

(JSC::StructureStubInfo::initializeFromUnlinkedStructureStubInfo):
(JSC::StructureStubInfo::initializeFromDFGUnlinkedStructureStubInfo):

• Source/JavaScriptCore/bytecode/StructureStubInfo.h:

(JSC::StructureStubInfo::inlineCodeSize const):
(JSC::StructureStubInfo::inlineSize const): Deleted.

• Source/JavaScriptCore/dfg/DFGInlineCacheWrapperInlines.h:

(JSC::DFG::InlineCacheWrapper<GeneratorType>::finalize):

• Source/JavaScriptCore/dfg/DFGJITCode.h:
• Source/JavaScriptCore/dfg/DFGOSRExitCompilerCommon.cpp:

(JSC::DFG::callerReturnPC):

• Source/JavaScriptCore/jit/JIT.cpp:

(JSC::JIT::link):

• Source/JavaScriptCore/jit/JITInlineCacheGenerator.cpp:

(JSC::JITInlineCacheGenerator::finalize):
(JSC::JITGetByValGenerator::generateEmptyPath):

• Source/JavaScriptCore/jit/JITInlineCacheGenerator.h:
• Source/JavaScriptCore/jit/JITPropertyAccess.cpp:

(JSC::JIT::emit_op_get_by_val):

• JSTests/stress/get-by-val-generic-structurestubinfo.js: Added.

(let.program):
(runMono.let.o.get x):
(runMono):
(runPoly):

Canonical link: ​https://commits.webkit.org/251619@main

2:48 PM Changeset in webkit [295613] by eric.carlson@apple.com

• 1 edit in trunk/Source/WebCore/platform/mediastream/mac/ScreenCaptureKitCaptureSource.mm

[macOS] Unable to resume screen sharing after it is paused
​https://bugs.webkit.org/show_bug.cgi?id=241693
rdar://93573937

Reviewed by Youenn Fablet.

• Source/WebCore/platform/mediastream/mac/ScreenCaptureKitCaptureSource.mm:

(WebCore::ScreenCaptureKitCaptureSource::stop): Clear m_contentStream in the completion handler.
(WebCore::ScreenCaptureKitCaptureSource::startContentStream): Don't recreate instance variables
that already exist.

Canonical link: ​https://commits.webkit.org/251618@main

1:45 PM Changeset in webkit [295612] by Jonathan Bedard

• 6 edits in trunk/Tools/Scripts/libraries

[webkitcorepy] Validate cached credentials
​https://bugs.webkit.org/show_bug.cgi?id=241664
<rdar://problem/95254368>

Reviewed by Aakash Jain.

• Tools/Scripts/libraries/webkitbugspy/setup.py: Bumpv version.
• Tools/Scripts/libraries/webkitbugspy/webkitbugspy/init.py: Ditto.
• Tools/Scripts/libraries/webkitbugspy/webkitbugspy/mocks/github.py: Add user endpoint.
• Tools/Scripts/libraries/webkitcorepy/setup.py: Bump version.
• Tools/Scripts/libraries/webkitcorepy/webkitcorepy/init.py: Ditto.
• Tools/Scripts/libraries/webkitcorepy/webkitcorepy/credentials.py:

(credentials): Validate cached credentials, do not reload credentials which fail validation
from the keychain.

Canonical link: ​https://commits.webkit.org/251617@main

1:41 PM Changeset in webkit [295611] by Dewei Zhu

• 4 edits in trunk/Websites/perf.webkit.org

Prefer using commit identifier to build commit URL.
​https://bugs.webkit.org/show_bug.cgi?id=241646
rdar://93104485

Reviewed by Jonathan Bedard.

Use commit identifier to construct commit URL and blame URL when available.
This simplifies URL construction for Git transition.

• Websites/perf.webkit.org/public/v3/models/commit-log.js: Added code to prefer using commit identifiers

when available.
(CommitLog.prototype.url):
(CommitLog.prototype.diff):

• Websites/perf.webkit.org/unit-tests/commit-log-tests.js: Added a unit test for diff.

(assert.deepStrictEqual.webkitGitCommitWithRevisionIdentifier.diff.oldWebKitGitCommitWithRevisionIdentifier):

• Websites/perf.webkit.org/unit-tests/commit-set-tests.js: Updated unit test.
• Websites/perf.webkit.org/unit-tests/resources/mock-v3-models.js: Added 'blameUrl' to 'WebKitGit' mock repository.

(MockModels.inject):

Canonical link: ​https://commits.webkit.org/251616@main

1:37 PM Changeset in webkit [295610] by Alexey Shvayka

• 1 edit in trunk/Source/JavaScriptCore/bytecode/PropertyCondition.cpp

AbsenceOfSetEffect property condition should mind put() overrides
​https://bugs.webkit.org/show_bug.cgi?id=241574
<rdar://91833733>

Reviewed by Yusuke Suzuki.

Since JSArray's "length" and RegExpObject's "lastIndex" may be reconfigured as non-writable,
we need to handle them separately in AbsenceOfSetEffect property condition to ensure that compiler
takes a slow path in that case, following the spec and throwing an exception in strict mode [1].

I'm not sure how to make a test case capturing this though.

[1]: ​https://tc39.es/ecma262/#sec-ordinarysetwithowndescriptor (step 2.a)

• Source/JavaScriptCore/bytecode/PropertyCondition.cpp:

(JSC::nonStructurePropertyMayBecomeReadOnlyWithoutTransition):
(JSC::PropertyCondition::isStillValidAssumingImpurePropertyWatchpoint const):

Canonical link: ​https://commits.webkit.org/251615@main

1:33 PM Changeset in webkit [295609] by Devin Rousso

• 2 edits in trunk/Source/WebInspectorUI

Web Inspector: use proper name for ITP debug mode
​https://bugs.webkit.org/show_bug.cgi?id=241697

Reviewed by Patrick Angle.

• Source/WebInspectorUI/UserInterface/Base/Main.js:

(WI._handleDeviceSettingsTabBarButtonClicked):

• Source/WebInspectorUI/Localizations/en.lproj/localizedStrings.js:

Canonical link: ​https://commits.webkit.org/251614@main

1:11 PM Changeset in webkit [295608] by Alexey Shvayka

• 8 edits
  1 add in trunk

CommonSlowPaths::putDirectWithReify() is incorrect for DontDelete properties
​https://bugs.webkit.org/show_bug.cgi?id=241651
<rdar://94016559>

Reviewed by Saam Barati.

Provided the base object has no read-only / accessor / custom properties,
putDirectWithReify() is incorrect for an object with non-configurable property,
whether it's on the structure or a non-reified static none.

In that case, putDirectWithReify() ignores existing non-configurable / non-reified
descriptor and produces an incorrect property descriptor instead of throwing TypeError.
One it's observed in the wild is via an instance field [1].

The issue was due to incorrect ReadOnly-focused check for putDirect() fast path,
which would be correct for Set but not for DefineOwnProperty.

Apart from introducing tighter fast patch check, this change extracts
JSFunction::mayHaveNonReifiedPrototype() helper, cleaning up JSFunction's overrides,
and removes now unused PutModeDefineOwnPropertyIgnoringExtensibility, which apart from
ignoring extensibility, also ignored read-only / accessor / custom properties,
which felt a bit counter-intuitive.

This change carefully preserves the fast path introduced in webkit.org/b/232479.

[1]: ​https://tc39.es/ecma262/#sec-definefield

• Source/JavaScriptCore/runtime/CommonSlowPaths.h:

(JSC::CommonSlowPaths::originalStructureBeforePut):
(JSC::CommonSlowPaths::canPutDirectFast):
(JSC::CommonSlowPaths::putDirectWithReify):
(JSC::CommonSlowPaths::putDirectAccessorWithReify):

• Source/JavaScriptCore/runtime/JSFunction.cpp:

(JSC::JSFunction::getOwnPropertySlot):
(JSC::JSFunction::put):
(JSC::JSFunction::deleteProperty):
(JSC::JSFunction::defineOwnProperty):
(JSC::JSFunction::reifyLazyPropertyIfNeeded):
(JSC::JSFunction::reifyLazyPrototypeIfNeeded):

• Source/JavaScriptCore/runtime/JSFunction.h:
• Source/JavaScriptCore/runtime/JSFunctionInlines.h:

(JSC::JSFunction::mayHaveNonReifiedPrototype):

• Source/JavaScriptCore/runtime/JSObject.cpp:

(JSC::JSObject::putDirectCustomAccessor):
(JSC::JSObject::putDirectNonIndexAccessor):

• Source/JavaScriptCore/runtime/JSObject.h:

(JSC::JSObject::putDirect):
(JSC::JSObject::putDirectRespectingExtensibility): Deleted.

• Source/JavaScriptCore/runtime/JSObjectInlines.h:

(JSC::JSObject::putDirectInternal):

• Source/JavaScriptCore/tools/JSDollarVM.cpp:

(JSC::JSC_DEFINE_HOST_FUNCTION):
(JSC::JSDollarVM::finishCreation):

Canonical link: ​https://commits.webkit.org/251613@main

1:00 PM Changeset in webkit [295607] by Megan Gardner

• 2 edits in trunk/Source

Implement alternate fullscreen controls.
​https://bugs.webkit.org/show_bug.cgi?id=241654

Reviewed by Tim Horton.

Alternative UI for fullscreen video controls.

• Source/WTF/Scripts/Preferences/WebPreferences.yaml:
• Source/WebKit/UIProcess/ios/fullscreen/WKFullScreenViewController.mm:

(-[WKFullScreenViewController loadView]):

Canonical link: ​https://commits.webkit.org/251612@main

12:51 PM Changeset in webkit [295606] by sbarati@apple.com

• 1 edit in trunk/Source/JavaScriptCore/b3/air/AirAllocateRegistersAndStackAndGenerateCode.cpp

Change how we rewind instructions in Air's O0 register allocator
​https://bugs.webkit.org/show_bug.cgi?id=241687

Reviewed by Mark Lam.

• Source/JavaScriptCore/b3/air/AirAllocateRegistersAndStackAndGenerateCode.cpp:

(JSC::B3::Air::GenerateAndAllocateRegisters::generate):

Canonical link: ​https://commits.webkit.org/251611@main

12:44 PM Changeset in webkit [295605] by Elliott Williams

• 6 edits in trunk

Add lldbWebKitTester to WebKit.xcworkspace

Unreviewed test fix.

When we switched build-webkit to workspace builds in
​https://commits.webkit.org/251501@main, we stopped building it.

lldbWebKitTester must only build on macOS. Add a "lldbWebKitTester
(Platform filters)" aggregate target that selectively builds
lldbWebKitTester on Mac only, like we did for MiniBrowser in
​https://commits.webkit.org/251501@main. Update the relevant schemes to
build that aggregate.

We cannot rely on lldbWebKitTester's "SUPPORTED_PLATFORMS = macosx" as a
way to make it always build for Mac, because it depends on WTF and
therefore must only build if WTF is also being built for Mac.

• Tools/lldb/lldbWebKitTester/Configurations/Base.xcconfig: Claim to support every platform, so that the aggregate target is always buildable and able to apply a platform filter.
• Tools/lldb/lldbWebKitTester/Configurations/lldbWebKitTester.xcconfig: Move "SUPPORTED_PLATFORMS = macosx" here, because lldbWebKitTester really does need to be restricted to building for Mac.

• Tools/lldb/lldbWebKitTester/lldbWebKitTester.xcodeproj/project.pbxproj:
• WebKit.xcworkspace/xcshareddata/xcschemes/All Modules.xcscheme:
• WebKit.xcworkspace/xcshareddata/xcschemes/All Tools.xcscheme:

Canonical link: ​https://commits.webkit.org/251610@main

12:33 PM Changeset in webkit [295604] by Kate Cheney

• 1 edit in trunk/Source/WebCore/platform/graphics/coretext/DrawGlyphsRecorderCoreText.cpp

File size label for attachment in mail compose shows up black instead of expected grey
​https://bugs.webkit.org/show_bug.cgi?id=241600
rdar://94071484

Reviewed by Cameron McCormack.

As of ​https://trac.webkit.org/changeset/293867/webkit in DrawGlyphsRecorder
specifically, we use the fill color of the initial state of GraphicsContext
in the case where the new fill color we are updating to is equal to the existing
fill color for perf reasons. This revealed a bug where CGContext state gets
out of sync with it's wrapping GraphicsContext, causing an incorrect
initial state fill color and thus painting the second line of Mail
attachments in the wrong color.

This patch fixes this bug by saving and restoring the state of
m_internalContext, which we reference in the case above to determine
whether to update the fill color.

• Source/WebCore/platform/graphics/coretext/DrawGlyphsRecorderCoreText.cpp:

(WebCore::DrawGlyphsRecorder::drawNativeText):

Canonical link: ​https://commits.webkit.org/251609@main

12:30 PM Changeset in webkit [295603] by commit-queue@webkit.org

• 5 edits in trunk/Source/WebCore/workers

SharedWorkerGlobalScope::close should not directly stop its thread
​https://bugs.webkit.org/show_bug.cgi?id=241677
rdar://95010255

Patch by Youenn Fablet <​youennf@gmail.com> on 2022-06-16
Reviewed by Alex Christensen.

We were directly stopping the worker thread when calling SharedWorkerGlobalScope::close.
This can trigger the case of stopping the worker thread concurrently from main thread and shared worker thread.
Instead, we reuse the existing Worker infastructure to tell its SharedWorkerThreadProxy to terminate.

• Source/WebCore/workers/WorkerOrWorkletThread.cpp:

(WebCore::WorkerOrWorkletThread::stop):

• Source/WebCore/workers/shared/SharedWorkerGlobalScope.cpp:

(WebCore::SharedWorkerGlobalScope::close): Deleted.

• Source/WebCore/workers/shared/SharedWorkerGlobalScope.h:
• Source/WebCore/workers/shared/context/SharedWorkerThreadProxy.cpp:

(WebCore::SharedWorkerThreadProxy::workerGlobalScopeClosed):

• Source/WebCore/workers/shared/context/SharedWorkerThreadProxy.h:

Canonical link: ​https://commits.webkit.org/251608@main

11:49 AM Changeset in webkit [295602] by Alexey Shvayka

• 12 edits
  5 adds in trunk

[WebIDL] maplike<> and setlike<> declarations should be resilient to tampered prototypes
​https://bugs.webkit.org/show_bug.cgi?id=241617
<rdar://93229569>

Reviewed by Yusuke Suzuki.

With this change, maplike<> and setlike<> declarations work as expected if methods of
Map.prototype / Set.prototype are removed, as they are suppossed to per spec [1][2].
Usage of backing Map / Set is an implementation detail of WebKit bindings and should
not be observable.

The fix mirrors all Map / Set prototype methods and "size" getter by private names,
which are inacessible to userland code, ensuring that public JSFunction* instances
are reused to avoid memory bloat.

Also, this change:

• saves creating 4 extra JSFunction* instances during init of Map / Set prototypes;
• speeds-up call forwarding by retrieving methods from prototypes with getDirect();
• aligns property order with the spec, even though there is no requirement.

[1] ​https://webidl.spec.whatwg.org/#es-maplike
[2] ​https://webidl.spec.whatwg.org/#es-setlike

• LayoutTests/imported/w3c/web-platform-tests/css/css-highlight-api/Highlight-setlike-tampered-Set-prototype-expected.txt: Added.
• LayoutTests/imported/w3c/web-platform-tests/css/css-highlight-api/Highlight-setlike-tampered-Set-prototype.html: Added.
• LayoutTests/imported/w3c/web-platform-tests/css/css-highlight-api/HighlightRegistry-maplike-tampered-Map-prototype-expected.txt: Added.
• LayoutTests/imported/w3c/web-platform-tests/css/css-highlight-api/HighlightRegistry-maplike-tampered-Map-prototype.html: Added.
• Source/JavaScriptCore/DerivedSources-output.xcfilelist:
• Source/JavaScriptCore/DerivedSources.make:
• Source/JavaScriptCore/JavaScriptCore.xcodeproj/project.pbxproj:
• Source/JavaScriptCore/builtins/BuiltinNames.h:
• Source/JavaScriptCore/runtime/MapPrototype.cpp:

(JSC::MapPrototype::finishCreation):

• Source/JavaScriptCore/runtime/SetPrototype.cpp:

(JSC::SetPrototype::finishCreation):

• Source/WebCore/bindings/js/JSDOMBindingInternals.js:

(forEachWrapper):

• Source/WebCore/bindings/js/JSDOMMapLike.cpp:

(WebCore::getBackingMap):
(WebCore::clearBackingMap):
(WebCore::setToBackingMap):
(WebCore::forwardFunctionCallToBackingMap):

• Source/WebCore/bindings/js/JSDOMMapLike.h:

(WebCore::forwardSizeToMapLike):
(WebCore::forwardEntriesToMapLike):
(WebCore::forwardKeysToMapLike):
(WebCore::forwardValuesToMapLike):
(WebCore::forwardClearToMapLike):
(WebCore::forwardGetToMapLike):
(WebCore::forwardHasToMapLike):
(WebCore::forwardSetToMapLike):
(WebCore::forwardDeleteToMapLike):

• Source/WebCore/bindings/js/JSDOMSetLike.cpp:

(WebCore::getBackingSet):
(WebCore::clearBackingSet):
(WebCore::addToBackingSet):
(WebCore::forwardFunctionCallToBackingSet):

• Source/WebCore/bindings/js/JSDOMSetLike.h:

(WebCore::forwardSizeToSetLike):
(WebCore::forwardEntriesToSetLike):
(WebCore::forwardKeysToSetLike):
(WebCore::forwardValuesToSetLike):
(WebCore::forwardClearToSetLike):
(WebCore::forwardHasToSetLike):
(WebCore::forwardAddToSetLike):
(WebCore::forwardDeleteToSetLike):

• Source/WebCore/bindings/js/WebCoreBuiltinNames.h:

Canonical link: ​https://commits.webkit.org/251607@main

11:26 AM Changeset in webkit [295601] by timothy@apple.com

• 1 edit in trunk/metadata/contributors.json

Add xeenon GitHub account to contributors.json.
​https://bugs.webkit.org/show_bug.cgi?id=241686

Unreviewed.

• metadata/contributors.json:

Add xeenon GitHub account under Timothy Hatcher.

Canonical link: ​https://commits.webkit.org/251606@main

11:08 AM Changeset in webkit [295600] by pvollan@apple.com

• 1 edit in trunk/Source/WebKit/UIProcess/WebProcessPool.cpp

Handle display reconfiguration on main thread
​https://bugs.webkit.org/show_bug.cgi?id=241683
<rdar://83417447>

Reviewed by Simon Fraser.

It is not safe to handle this on a non main thread, since we send messages to the WebContent and GPU process.

• Source/WebKit/UIProcess/WebProcessPool.cpp:

(WebKit::displayReconfigurationCallBack):

Canonical link: ​https://commits.webkit.org/251605@main

11:06 AM Changeset in webkit [295599] by Jonathan Bedard

• 1 edit in trunk/Tools/CISupport/ews-build/steps.py

[Merge-Queue] Validate reviewers in commit message (Follow-up)
​https://bugs.webkit.org/show_bug.cgi?id=240718
<rdar://problem/93665771>

Unreviewed infrastructure fix.

• Tools/CISupport/ews-build/steps.py:

(ValidateCommitMessage.run): Allow for unreviewed changes.

10:32 AM Changeset in webkit [295598] by commit-queue@webkit.org

• 1 edit in trunk/Tools/Scripts/build-and-collect-pgo-profiles

Extend build-and-collect-pgo-profiles to support testing a browser path
​https://bugs.webkit.org/show_bug.cgi?id=241392
rdar://94581810

Patch by briannafan <​briannaf@berkeley.edu> on 2022-06-16
Reviewed by Dewei Zhu.

Extending script to take browser path and skip building.

• Tools/Scripts/build-and-collect-pgo-profiles:

Canonical link: ​https://commits.webkit.org/251603@main

9:39 AM Changeset in webkit [295597] by Tyler Wilcock

• 7 edits in trunk

AX: Update the isolated tree in response to dynamic placeholder and aria-placeholder changes
​https://bugs.webkit.org/show_bug.cgi?id=241675

Reviewed by Chris Fleizach.

• LayoutTests/accessibility/placeholder-expected.txt:
• LayoutTests/accessibility/placeholder.html: Add new test cases.
• LayoutTests/platform/ios/TestExpectations: Enable accessibility/placeholder.html.
• Source/WebCore/accessibility/AXLogger.cpp:

(WebCore::operator<<):

• Source/WebCore/accessibility/AXObjectCache.cpp:

(WebCore::AXObjectCache::handleAttributeChange):
(WebCore::AXObjectCache::updateIsolatedTree):

• Source/WebCore/accessibility/AXObjectCache.h:
• Source/WebCore/accessibility/atspi/AXObjectCacheAtspi.cpp

Canonical link: ​https://commits.webkit.org/251602@main

9:30 AM Changeset in webkit [295596] by sihui_liu@apple.com

• 1 edit in trunk/Source/WebKit/UIProcess/WebsiteData/WebsiteDataStore.cpp

Regression (r295056): do not delete WebSQL directory until there is no use of it in WebKit clients
​https://bugs.webkit.org/show_bug.cgi?id=241671

Reviewed by Youenn Fablet.

We have internal clients that set the custom WebSQL directory do not only use the directory for WebSQL.

• Source/WebKit/UIProcess/WebsiteData/WebsiteDataStore.cpp:

(WebKit::WebsiteDataStore::resolveDirectoriesIfNecessary):

Canonical link: ​https://commits.webkit.org/251601@main

9:27 AM Changeset in webkit [295595] by Claudio Saavedra

• 3 edits in trunk/Source/WebCore/platform/graphics/texmap

TextureMapper: remove unused size variable in draw texture methods
​https://bugs.webkit.org/show_bug.cgi?id=241679

Reviewed by Darin Adler.

WebCore::TextureMapperGL::drawTexturedQuadWithProgram() doesn't really
use the size parameter, remove it from everwhere where it's passed
forward.

• Source/WebCore/platform/graphics/texmap/TextureMapperGL.cpp:

(WebCore::TextureMapperGL::drawTexture):
(WebCore::prepareTransformationMatrixWithFlags):
(WebCore::TextureMapperGL::drawTexturePlanarYUV):
(WebCore::TextureMapperGL::drawTextureSemiPlanarYUV):
(WebCore::TextureMapperGL::drawTexturePackedYUV):
(WebCore::TextureMapperGL::drawTexturedQuadWithProgram):
(WebCore::TextureMapperGL::drawFiltered):
(WebCore::TextureMapperGL::drawTextureExternalOES):

• Source/WebCore/platform/graphics/texmap/TextureMapperGL.h:
• Source/WebCore/platform/graphics/texmap/TextureMapperPlatformLayerBuffer.cpp:

(WebCore::TextureMapperPlatformLayerBuffer::paintToTextureMapper):

Canonical link: ​https://commits.webkit.org/251600@main

8:22 AM Changeset in webkit [295594] by commit-queue@webkit.org

• 11 edits in trunk/Source

Tab snapshotting should not trigger the WebGL code path when HAVE(LOW_AV_SAMPLE_BUFFER_PRUNING_INTERVAL) is false
​https://bugs.webkit.org/show_bug.cgi?id=241628
rdar://94484791

Patch by Youenn Fablet <​youennf@gmail.com> on 2022-06-16
Reviewed by Jer Noble.

If HAVE(LOW_AV_SAMPLE_BUFFER_PRUNING_INTERVAL) is false, we either render the video or we can grab individual video frames for canvas rendering.
We render video except when WebGL canvas painting is used.
Before the patch, the GPUProcess code path was relying on calling MediaPlayerPrivateMediaSourceAVFObjC::videoFrameForCurrentTime for tab snapshotting.
This then prevented the video to be rendered.

This patch is adding a MediaPlayer::willBeAskedToPaintGL method to ask the player to enter the WebGL code path to grab individual video frames.
This method is called when creating a video texture for WebGL.
MediaPlayerPrivateMediaSourceAVFObjC implements this method to switch to the decompression session to grab individual video frames.
Other players are left unchanged.
We add the necessary IPC handling to send the signal from WebProcess to GPUProcess.
All of this is specific to HAVE(LOW_AV_SAMPLE_BUFFER_PRUNING_INTERVAL) being false.

Manually tested by tabsnapshooting youtube pages as well as loading web page using MSE+WebGL.

• Source/WebCore/html/canvas/WebGLRenderingContextBase.cpp:

(WebCore::WebGLRenderingContextBase::texImageSourceHelper):

• Source/WebCore/platform/graphics/MediaPlayer.cpp:

(WebCore::MediaPlayer::willBeAskedToPaintGL):

• Source/WebCore/platform/graphics/MediaPlayer.h:
• Source/WebCore/platform/graphics/MediaPlayerPrivate.h:

(WebCore::MediaPlayerPrivateInterface::willBeAskedToPaintGL):

• Source/WebCore/platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.h:
• Source/WebCore/platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.mm:

(WebCore::MediaPlayerPrivateMediaSourceAVFObjC::willBeAskedToPaintGL):
(WebCore::MediaPlayerPrivateMediaSourceAVFObjC::videoFrameForCurrentTime):
(WebCore::MediaPlayerPrivateMediaSourceAVFObjC::shouldEnsureLayer const):
(WebCore::MediaPlayerPrivateMediaSourceAVFObjC::acceleratedRenderingStateChanged):

• Source/WebKit/GPUProcess/media/RemoteMediaPlayerProxy.h:
• Source/WebKit/GPUProcess/media/RemoteMediaPlayerProxy.messages.in:
• Source/WebKit/GPUProcess/media/cocoa/RemoteMediaPlayerProxyCocoa.mm:

(WebKit::RemoteMediaPlayerProxy::willBeAskedToPaintGL):

• Source/WebKit/WebProcess/GPU/media/MediaPlayerPrivateRemote.cpp:

(WebKit::MediaPlayerPrivateRemote::willBeAskedToPaintGL):

• Source/WebKit/WebProcess/GPU/media/MediaPlayerPrivateRemote.h:

Canonical link: ​https://commits.webkit.org/251599@main

8:05 AM Changeset in webkit [295593] by Andres Gonzalez

• 1 edit in trunk/Source/WebCore/accessibility/mac/WebAccessibilityObjectWrapperMac.mm

AX ITM: VoiceOver cannot get contextual menus for links on some pages with ITM enabled.
​https://bugs.webkit.org/show_bug.cgi?id=241668

Reviewed by Chris Fleizach.

In isolated tree mode the performSelector method was being called off the main thread, and was not invoking the given selector. This patch fixes the problem by dispatching to the main thread the call to performSelector. Some code cleanup.

• Source/WebCore/accessibility/mac/WebAccessibilityObjectWrapperMac.mm:

(-[WebAccessibilityObjectWrapper accessibilityPerformShowMenuAction]):
(-[WebAccessibilityObjectWrapper _accessibilityShowContextMenu]):
(-[WebAccessibilityObjectWrapper accessibilityShowContextMenu]): Deleted.

Canonical link: ​https://commits.webkit.org/251598@main

7:46 AM Changeset in webkit [295592] by Jonathan Bedard

• 1 edit in trunk/Tools/CISupport/ews-build/steps.py

[ews-build.webkit.org] Support WebKit/WebKit-security
​https://bugs.webkit.org/show_bug.cgi?id=241656
<rdar://problem/95252686>

Reviewed by Ryan Haddad and Aakash Jain.

• Tools/CISupport/ews-build/steps.py:

Canonical link: ​https://commits.webkit.org/251597@main

7:37 AM Changeset in webkit [295591] by commit-queue@webkit.org

• 1 edit
  2 adds in trunk

A black screen appears in a muted video element outside the viewport
​https://bugs.webkit.org/show_bug.cgi?id=241152
rdar://problem/94562636

Patch by Youenn Fablet <​youennf@gmail.com> on 2022-06-16
Reviewed by Eric Carlson.

We sometimes remove the invisible autoplay restrictions.
If we remove it while we are suspended due this restriction, we will not unsuspend and will wait for a user interaction.
To prevent this, we do not return early in updateShouldAutoplay if we are interrupted due to the invisible autoplay restriction.

• LayoutTests/fast/mediastream/video-mediastream-restricted-invisible-autoplay-user-click-expected.txt: Added.
• LayoutTests/fast/mediastream/video-mediastream-restricted-invisible-autoplay-user-click.html: Added.
• Source/WebCore/html/HTMLMediaElement.cpp:

(WebCore::HTMLMediaElement::updateShouldAutoplay):

Canonical link: ​https://commits.webkit.org/251596@main

4:53 AM Changeset in webkit [295590] by commit-queue@webkit.org

• 1 edit
  3 adds in trunk

Moving a media element from one document to another should not break autoplay
​https://bugs.webkit.org/show_bug.cgi?id=241631
rdar://95060381

Patch by Youenn Fablet <​youennf@gmail.com> on 2022-06-16
Reviewed by Eric Carlson.

When pausing a video element when being detached from a document, autoplay might get broken if the element is interrupted due to invisibility.
In that case, the session will store the fact that the element is paused and when the end of invisibility interruption happens, the session state is Paused instead of Autoplay.
To prevent this, we do not pause when being detached if we are alread interrupted due to invisibility.

Covered by added test.

• LayoutTests/http/tests/webrtc/resources/utility-frame.html: Added.
• LayoutTests/http/tests/webrtc/video-mediastream-invisible-autoplay-detached-expected.txt: Added.
• LayoutTests/http/tests/webrtc/video-mediastream-invisible-autoplay-detached.html: Added.
• Source/WebCore/html/HTMLMediaElement.cpp:

(WebCore::HTMLMediaElement::pauseAfterDetachedTask):

Canonical link: ​https://commits.webkit.org/251595@main

1:07 AM Changeset in webkit [295589] by commit-queue@webkit.org

• 1 edit in trunk/Source/WebKit/NetworkProcess/cocoa/NetworkSessionCocoa.mm

Add WebSocket map size logging
​https://bugs.webkit.org/show_bug.cgi?id=241590
rdar://problem/95093425

Patch by Youenn Fablet <​youennf@gmail.com> on 2022-06-16
Reviewed by Eric Carlson.

• Source/WebKit/NetworkProcess/cocoa/NetworkSessionCocoa.mm:

(WebKit::NetworkSessionCocoa::addWebSocketTask):
(WebKit::NetworkSessionCocoa::removeWebSocketTask):

Canonical link: ​https://commits.webkit.org/251594@main

12:20 AM Changeset in webkit [295588] by achristensen@apple.com

• 1 edit in trunk/Source/WebCore/Modules/speech/SpeechRecognitionCaptureSourceImpl.cpp

TestWebKitAPI.WebKit2.SpeechRecognitionErrorWhenStartingAudioCaptureOnDifferentPageE is a flaky crash
​https://bugs.webkit.org/show_bug.cgi?id=239474

Reviewed by Sihui Liu.

At least when using MockAudioSharedInternalUnit the call to RealtimeMediaSource::start can call code on another thread
that assumes that RealtimeMediaSource::setLogger has already been called. This can be made to crash almost 100% of the time
in the test by adding WTFReportBacktrace at the beginning of RealtimeMediaSource::setLogger which delays when m_logger is set
enough for the "MockAudioSharedInternalUnit Capture Queue" thread to read from m_logger before it is written without this change.

• Source/WebCore/Modules/speech/SpeechRecognitionCaptureSourceImpl.cpp:

(WebCore::SpeechRecognitionCaptureSourceImpl::SpeechRecognitionCaptureSourceImpl):

Canonical link: ​https://commits.webkit.org/251593@main

Jun 15, 2022:

11:04 PM Changeset in webkit [295587] by Fujii Hironori

• 4 edits in trunk/Source/WebCore/platform/graphics/texmap

Remove unused code for GL_TEXTURE_RECTANGLE_ARB
​https://bugs.webkit.org/show_bug.cgi?id=241670

Reviewed by Don Olmstead.

The code for GL_TEXTURE_RECTANGLE_ARB was added by 105925@main (Bug
87738) for Qt-Mac port. But, no longer used.

• Source/WebCore/platform/graphics/texmap/TextureMapperGL.cpp:

(WebCore::TextureMapperGL::drawTexture):
(WebCore::prepareTransformationMatrixWithFlags):
(WebCore::TextureMapperGL::drawTexturePlanarYUV):
(WebCore::TextureMapperGL::drawTextureSemiPlanarYUV):
(WebCore::TextureMapperGL::drawTexturePackedYUV):
(WebCore::TextureMapperGL::drawTexturedQuadWithProgram):

• Source/WebCore/platform/graphics/texmap/TextureMapperGL.h:
• Source/WebCore/platform/graphics/texmap/TextureMapperShaderProgram.cpp:

(WebCore::TextureMapperShaderProgram::create):

• Source/WebCore/platform/graphics/texmap/TextureMapperShaderProgram.h:

Canonical link: ​https://commits.webkit.org/251592@main

10:49 PM Changeset in webkit [295586] by msaboff@apple.com

• 2 edits in trunk/Source/WebKit/Configurations

[Regression r295565] iOS Simulator build fails
​https://bugs.webkit.org/show_bug.cgi?id=241673

Unreviewed, fix ios simulator build after r295565.

• Source/WebKit/Configurations/adattributiond.xcconfig:
• Source/WebKit/Configurations/webpushd.xcconfig:

Canonical link: ​https://commits.webkit.org/251591@main

10:31 PM Changeset in webkit [295585] by sbarati@apple.com

• 5 edits in trunk/Source/JavaScriptCore

Don't clear both prototype ICs for LLInt iterator_next
​https://bugs.webkit.org/show_bug.cgi?id=241663

Reviewed by Yusuke Suzuki.

If we use BytecodeIndex instead of a bytecode offset, we don't need to
clear both ICs. We can just clear the one that the watchpoint fired for.

• Source/JavaScriptCore/bytecode/CodeBlock.cpp:

(JSC::CodeBlock::finalizeLLIntInlineCaches):

• Source/JavaScriptCore/bytecode/CodeBlock.h:
• Source/JavaScriptCore/bytecode/LLIntPrototypeLoadAdaptiveStructureWatchpoint.cpp:

(JSC::LLIntPrototypeLoadAdaptiveStructureWatchpoint::LLIntPrototypeLoadAdaptiveStructureWatchpoint):
(JSC::LLIntPrototypeLoadAdaptiveStructureWatchpoint::initialize):
(JSC::LLIntPrototypeLoadAdaptiveStructureWatchpoint::fireInternal):

• Source/JavaScriptCore/bytecode/LLIntPrototypeLoadAdaptiveStructureWatchpoint.h:
• Source/JavaScriptCore/llint/LLIntSlowPaths.cpp:

(JSC::LLInt::setupGetByIdPrototypeCache):
(JSC::LLInt::performLLIntGetByID):
(JSC::LLInt::LLINT_SLOW_PATH_DECL):

Canonical link: ​https://commits.webkit.org/251590@main

9:57 PM Changeset in webkit [295584] by Fujii Hironori

• 1 edit in trunk/Source/WebCore/platform/graphics/texmap/TextureMapperShaderProgram.h

Remove unused expandedQuadEdgesInScreenSpace uniform variable
​https://bugs.webkit.org/show_bug.cgi?id=241667

Reviewed by Don Olmstead.

expandedQuadEdgesInScreenSpace uniform variable was added by
108289@main (Bug 90308), but unsued since 124039@main (Bug 105377).

• Source/WebCore/platform/graphics/texmap/TextureMapperShaderProgram.h:

Removed it.

Canonical link: ​https://commits.webkit.org/251589@main

9:16 PM Changeset in webkit [295583] by commit-queue@webkit.org

• 2 edits in trunk/Source/ThirdParty/ANGLE

Validate descriptor when creating MTLRenderPipelineState
​https://bugs.webkit.org/show_bug.cgi?id=241587
rdar://problem/93820440

Patch by Dan Glastonbury <​djg@apple.com> on 2022-06-15
Reviewed by Kimmo Kinnunen.

• Source/ThirdParty/ANGLE/src/libANGLE/renderer/metal/mtl_state_cache.mm:

(rx::mtl::ValidateRenderPipelineState):
(rx::mtl::RenderPipelineCache::createRenderPipelineState):
Extend MTlRenderPipelineDescriptor validation to ensure that there is at least
one valid render target set for the the render pipeline. This is required for
certain families of metal devices to avoid a validation failure inside the metal
framework. Moving the failure here will cause the app using ANGLE to return a GL
error instead of crashing the process.

Canonical link: ​https://commits.webkit.org/251588@main

7:18 PM Changeset in webkit [295582] by Andres Gonzalez

• 1 edit in trunk/Source/WebCore/accessibility/isolatedtree/AXIsolatedTree.cpp

AX ITM: Should not build an isolated tree branch rooted at an object with invalid ID. Should not attempt to remove an object for an invalid ID.
​https://bugs.webkit.org/show_bug.cgi?id=241643

Reviewed by Chris Fleizach.

Added check for valid object ID in AXIsolatedTree::collectNodeChangesForSubtree and in updateChildren since in both cases the ID is used as a key for HashMaps which cannot be null.

• Source/WebCore/accessibility/isolatedtree/AXIsolatedTree.cpp:

(WebCore::AXIsolatedTree::collectNodeChangesForSubtree):
(WebCore::AXIsolatedTree::updateChildren):

Canonical link: ​https://commits.webkit.org/251587@main

6:02 PM Changeset in webkit [295581] by commit-queue@webkit.org

• 1 edit in trunk/Source/cmake/WebKitCommon.cmake

[CMake] Check for all required perl modules
​https://bugs.webkit.org/show_bug.cgi?id=241648

Patch by Michael Catanzaro <​mcatanzaro@redhat.com> on 2022-06-15
Reviewed by Adrian Perez de Castro.

• Source/cmake/WebKitCommon.cmake:

Canonical link: ​https://commits.webkit.org/251586@main

4:55 PM Changeset in webkit [295580] by Patrick Angle

• 3 edits in trunk/LayoutTests

Web Inspector: inspector/page/setShowPaintRects.html is a flakey failure since introduction
​https://bugs.webkit.org/show_bug.cgi?id=233047
rdar://85345706

Reviewed by Antoine Quint.

Adding test results to the page causes a paint to be scheduled, which can cause paint rects to be requested at times we
are not expecting them as part of the flow of the test. We expect only our explicit change to an element to cause paints
to occur, but currently those paints end up interleaved with paints for adding results to the test page.

To resolve this, we now defer writing results to the test page until the end of each test case. This allows us to wait
to start each test case until there are no paint rects, and to avoid accidently causing a paint rect to appear at
inopportune times for our testing.

• LayoutTests/inspector/page/setShowPaintRects-expected.txt:
• LayoutTests/inspector/page/setShowPaintRects.html:
• LayoutTests/platform/mac/TestExpectations:

Canonical link: ​https://commits.webkit.org/251585@main

4:32 PM Changeset in webkit [295579] by Patrick Angle

• 1 edit in trunk/.github/CODEOWNERS

Add myself (patrickangle) to GitHub CODEOWNERS for Web Inspector

Reviewed by Jonathan Bedard.

• .github/CODEOWNERS:

Canonical link: ​https://commits.webkit.org/251584@main

4:01 PM Changeset in webkit [295578] by commit-queue@webkit.org

• 2 edits in trunk/Source/WTF

[Cocoa] Add feature flag for WebM Experiment
​https://bugs.webkit.org/show_bug.cgi?id=241642
<rdar://95228097>

Patch by Youssef Soliman <​youssefdevelops@gmail.com> on 2022-06-15
Reviewed by Eric Carlson.

Added internal feature and build flag for experimental WebM player.

• Source/WTF/Scripts/Preferences/WebPreferencesInternal.yaml:
• Source/WTF/wtf/PlatformEnableCocoa.h:

Canonical link: ​https://commits.webkit.org/251583@main

3:46 PM Changeset in webkit [295577] by Elliott Williams

• 1 edit in trunk/Tools/Scripts/build-webkit

[build-webkit] Don't run build-imagediff when cleaning with Xcode
​https://bugs.webkit.org/show_bug.cgi?id=241624

Reviewed by Jonathan Bedard.

Fixes "error: Unknown build action ' --clean'" from xcodebuild, as
build-webkit was passing the --clean argument to build-imagediff with an
extra space.

Since the move to XCBuild, cleaning deletes the whole build directory,
so there is nothing left for build-imagediff to clean and we skip this
step altogether.

• Tools/Scripts/build-webkit:

Canonical link: ​https://commits.webkit.org/251581@main

3:46 PM Changeset in webkit [295576] by ysuzuki@apple.com

• 1 edit in trunk/.github/CODEOWNERS

Add CODEOWNERS file to ping review request automatically to JSC reviewers
​https://bugs.webkit.org/show_bug.cgi?id=241058

Reviewed by Jonathan Bedard.

This patch adds JavaScriptCore CODEOWNERS configurations. By using this file, we can request
reviews automatically based on modified files. I added jsc-reviewers group and
this file configures that Source/JavaScriptCore (except for inspector directory)
PR will automatically set jsc-reviewers as a requested reviewer.

• .github/CODEOWNERS: Added.

Canonical link: ​https://commits.webkit.org/251581@main

3:29 PM Changeset in webkit [295575] by Devin Rousso

• 1 edit in trunk/.github/CODEOWNERS

add myself to GitHub CODEOWNERS

Reviewed by Jonathan Bedard

• .github/CODEOWNERS:

Specifically for:

• Web Inspector (frontend and backend)
• Apple Pay
• Payment Request
• media controls

Canonical link: ​https://commits.webkit.org/251580@main

3:21 PM Changeset in webkit [295574] by Patrick Angle

• 3 edits in trunk/Source/WebInspectorUI/UserInterface/Views

Web Inspector: Regression(r293334) fill-available is not a valid dimension, -webkit-fill-available is still prefixed
​https://bugs.webkit.org/show_bug.cgi?id=241649
rdar://95240065

Reviewed by Devin Rousso.

fill-available is not a valid CSS dimension, we still need to use -webkit-fill-available for now.

• Source/WebInspectorUI/UserInterface/Views/EventBreakpointPopover.css:

(.popover .edit-breakpoint-popover-content .event input):

• Source/WebInspectorUI/UserInterface/Views/GeneralStyleDetailsSidebarPanel.css:

(.sidebar > .panel.details.css-style > .content ~ .options-container > .filter-bar):

• Source/WebInspectorUI/UserInterface/Views/RecordingContentView.css:

(.content-view.recording > .preview-container):

Canonical link: ​https://commits.webkit.org/251579@main

2:40 PM Changeset in webkit [295573] by Antti Koivisto

• 1 edit
  2 adds in trunk

REGRESSION (r287195): Safari fails to correctly render indented numbered lists with custom CSS and hyphenation ON
​https://bugs.webkit.org/show_bug.cgi?id=241630
<rdar://91245970>

Reviewed by Darin Adler.

• LayoutTests/fast/text/list-and-hyphenation-expected.html: Added.
• LayoutTests/fast/text/list-and-hyphenation.html: Added.
• Source/WebCore/rendering/RenderText.cpp:

(WebCore::RenderText::maxWordFragmentWidth):

Return entireWordWidth instead of 0 in the bail out cases.
This ends up setting maxWordWidth = w and leaving currMinWidth unchanged in the caller, matching the pre-r287195 behavior.

Canonical link: ​https://commits.webkit.org/251578@main

2:35 PM Changeset in webkit [295572] by pvollan@apple.com

• 2 edits in trunk/Source/WebKit

Block network extension system calls
​https://bugs.webkit.org/show_bug.cgi?id=241633
<rdar://73513138>

Reviewed by Sihui Liu.

After moving content filtering to the Network process and fixing an issue where Safari's injected bundle were
making network calls, network extension system calls can be blocked in the WebContent process.

• Source/WebKit/Resources/SandboxProfiles/ios/com.apple.WebKit.WebContent.sb.in:
• Source/WebKit/WebProcess/com.apple.WebProcess.sb.in:

Canonical link: ​https://commits.webkit.org/251577@main

1:44 PM Changeset in webkit [295571] by pvollan@apple.com

• 1 edit
  1 move in trunk/Source/WebKit

Rename file
​https://bugs.webkit.org/show_bug.cgi?id=241644

Unreviewed, rename incorrectly named file AuxiliaryProcessProcyCocoa.mm.

• Source/WebKit/UIProcess/Cocoa/AuxiliaryProcessProxyCocoa.mm: Renamed from Source/WebKit/UIProcess/Cocoa/AuxiliaryProcessProcyCocoa.mm.

(WebKit::AuxiliaryProcessProxy::fetchAudioComponentServerRegistrations):

• Source/WebKit/WebKit.xcodeproj/project.pbxproj:

Canonical link: ​https://commits.webkit.org/251576@main

1:20 PM Changeset in webkit [295570] by Brandon

• 2 edits in trunk/Source/bmalloc/libpas/src/libpas

[libpas] fix build with pas_assertion_failed and type conversion
​https://bugs.webkit.org/show_bug.cgi?id=241640

Reviewed by Mark Lam.

Build fix.

• Source/bmalloc/libpas/src/libpas/pas_utils.c:

(pas_report_assertion_failed):
(pas_assertion_failed): Deleted.

• Source/bmalloc/libpas/src/libpas/pas_utils.h:

(pas_assertion_failed):
(pas_assertion_failed_noreturn_silencer1):
(pas_assertion_failed_noreturn_silencer2):
(pas_assertion_failed_noreturn_silencer3):
(pas_assertion_failed_noreturn_silencer4):
(pas_assertion_failed_noreturn_silencer5):
(pas_assertion_failed_noreturn_silencer6):

Canonical link: ​https://commits.webkit.org/251575@main

12:59 PM Changeset in webkit [295569] by Antti Koivisto

• 11 edits in trunk

Invalidate animation keyframes using container units on when container size changes
​https://bugs.webkit.org/show_bug.cgi?id=241546

Reviewed by Antoine Quint.

Container size change also changes the interpretation of container units used in keyframes.

• LayoutTests/imported/w3c/web-platform-tests/css/css-contain/container-queries/container-units-animation-expected.txt:
• Source/WebCore/dom/Element.cpp:

(WebCore::Element::invalidateForQueryContainerSizeChange):
(WebCore::Element::needsUpdateQueryContainerDependentStyle const):
(WebCore::Element::clearNeedsUpdateQueryContainerDependentStyle):
(WebCore::Element::invalidateForQueryContainerChange): Deleted.

Add a new bit that tells when a container has been resized.

• Source/WebCore/dom/Element.h:
• Source/WebCore/dom/Node.h:
• Source/WebCore/rendering/style/KeyframeList.cpp:

(WebCore::KeyframeList::usesContainerUnits const):

Check for container unit use.

• Source/WebCore/rendering/style/KeyframeList.h:
• Source/WebCore/style/StyleScope.cpp:

(WebCore::Style::Scope::updateQueryContainerState):

• Source/WebCore/style/StyleTreeResolver.cpp:

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

Invalidate the keyframes if needed when computing the style.

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

Track if the subtree is withing a resized container.

• Source/WebCore/style/StyleTreeResolver.h:
• Source/WebCore/style/Styleable.cpp:

(WebCore::Styleable::queryContainerDidChange const):

• Source/WebCore/style/Styleable.h:

Canonical link: ​https://commits.webkit.org/251574@main

12:28 PM Changeset in webkit [295568] by eric.carlson@apple.com

• 1 edit in trunk/Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm

[Cocoa] Crash in SourceBufferPrivateAVFObjC::layerDidReceiveError.
​https://bugs.webkit.org/show_bug.cgi?id=241599
rdar://92972452

Reviewed by Jer Noble.

Have WebAVSampleBufferErrorListener take a strong reference to parent before calling it.

• Source/WebCore/platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:

(-[WebAVSampleBufferErrorListener observeValueForKeyPath:ofObject:change:context:]):
(-[WebAVSampleBufferErrorListener layerFailedToDecode:]):
(-[WebAVSampleBufferErrorListener audioRendererWasAutomaticallyFlushed:]):

Canonical link: ​https://commits.webkit.org/251573@main

11:01 AM Changeset in webkit [295567] by pvollan@apple.com

• 1 edit in trunk/Source/WebKit/Shared/Cocoa/AuxiliaryProcessCocoa.mm

Fix build issue
​https://bugs.webkit.org/show_bug.cgi?id=241637

Unreviewed, fix build issue after r295560.

• Source/WebKit/Shared/Cocoa/AuxiliaryProcessCocoa.mm:

Canonical link: ​https://commits.webkit.org/251572@main

10:58 AM Changeset in webkit [295566] by commit-queue@webkit.org

• 1 edit in trunk/Websites/bugs.webkit.org/PrettyPatch/PrettyPatch.rb

Incorrect Github file link in PrettyPatch
​https://bugs.webkit.org/show_bug.cgi?id=241638

Patch by Youssef Soliman <​youssefdevelops@gmail.com> on 2022-06-15
Reviewed by Jonathan Bedard.

In the pretty patch view, pressing a filename redirects you to
​https://github.com/WebKit/WebKit/blob/%7Bfilename%7D due to a
templating typo.

• Websites/bugs.webkit.org/PrettyPatch/PrettyPatch.rb:

Canonical link: ​https://commits.webkit.org/251571@main

10:51 AM Changeset in webkit [295565] by msaboff@apple.com

• 7 edits
  1 add in trunk/Source/WebKit

Move webpushd and adattributiond to the secondary content path when configured
​https://bugs.webkit.org/show_bug.cgi?id=241539
<rdar://90513797>

Reviewed by Keith Miller.

Moved webpushd and adattributiond and their plist files to the secondary content path when using
the system content path.
As part of this change, webpushd was moved to SYSTEM_SECONDARY_CONTENT_PATH/usr/libexec.
Created symlinks in /System/Library/{LaunchAgents|LaunchDaemons} for the plist files.

• Source/WebKit/Configurations/Base.xcconfig:
• Source/WebKit/Configurations/adattributiond.xcconfig:
• Source/WebKit/Configurations/webpushd.xcconfig:
• Source/WebKit/Scripts/copy-launchd-plist-and-create-symlink.sh: Added.
• Source/WebKit/Shared/EntryPointUtilities/Cocoa/Daemon/com.apple.webkit.adattributiond.plist:
• Source/WebKit/WebKit.xcodeproj/project.pbxproj:
• Source/WebKit/webpushd/com.apple.webkit.webpushd.ios.plist:
• Source/WebKit/webpushd/com.apple.webkit.webpushd.mac.plist:

Canonical link: ​https://commits.webkit.org/251570@main

10:06 AM Changeset in webkit [295564] by Aditya Keerthi

• 3 edits in trunk/LayoutTests

REGRESSION: [ Mac wk2 ] fast/forms/datetimelocal/datetimelocal-editable-components/datetimelocal-choose-value-from-picker.html is a flaky failure
​https://bugs.webkit.org/show_bug.cgi?id=239257
rdar://91653542

Reviewed by Wenson Hsieh.

245100@main changed the value parsing and sanitization algorithm for
<input type="datetime-local"> to better match the spec. Following this change,
the output of HTMLInputElement.value uses the shortest possible string,
omitting seconds or milliseconds when 0.

The flaky test tests that choosing a value using the date/time picker does not
unexpectedly add or remove second and millisecond fields. This behavior was
being verified by checking whether the input's value contained seconds or
milliseconds after using the picker. However, following 245100@main, the value
can appear without seconds/milliseconds, even though the editable fields do.

The non-determinism comes from the fact that the current system time is used
to populate the seconds and milliseconds field when using the picker on an
empty input. Consequently, the test will fail when the current time has a 0
second value.

To fix, verify that using the picker does not unexpectedly add or remove second
and millisecond fields by checking the width of the element before and after
date selection. This approach ensures determinism, while fulfilling the goal of
the test.

As an aside, the current behavior to populate the seconds and milliseconds for
empty inputs could be improved. ​http://webkit.org/b/241620 has been filed, with
the intent of setting the seconds and milliseconds fields to 0 when they are
initially empty.

• LayoutTests/fast/forms/datetimelocal/datetimelocal-editable-components/datetimelocal-choose-value-from-picker-expected.txt:
• LayoutTests/fast/forms/datetimelocal/datetimelocal-editable-components/datetimelocal-choose-value-from-picker.html:
• LayoutTests/platform/mac-wk2/TestExpectations:

Canonical link: ​https://commits.webkit.org/251569@main

9:59 AM Changeset in webkit [295563] by Aditya Keerthi

• 3 edits in trunk/Source/WebKit/WebProcess/WebPage

REGRESSION (iOS 16): Find on Page highlights shift after entering Split View
​https://bugs.webkit.org/show_bug.cgi?id=241615
rdar://94795621

Reviewed by Wenson Hsieh.

When entering Split View, the find overlay layer and the highlighted range
should be redrawn to reflect changes in layout.

• Source/WebKit/WebProcess/WebPage/WebFoundTextRangeController.cpp:

(WebKit::WebFoundTextRangeController::decorateTextRangeWithStyle):
(WebKit::WebFoundTextRangeController::redraw):
(WebKit::WebFoundTextRangeController::setTextIndicatorWithRange):

Factor out text indicator creation into a common method.

• Source/WebKit/WebProcess/WebPage/WebFoundTextRangeController.h:
• Source/WebKit/WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::mainFrameDidLayout):

Redraw decorated ranges after layout.

Canonical link: ​https://commits.webkit.org/251568@main

9:43 AM Changeset in webkit [295562] by mark.lam@apple.com

• 4 edits in trunk/Source/bmalloc/libpas/src/libpas

Add variadic PAS_ASSERT support.
​https://bugs.webkit.org/show_bug.cgi?id=241626
rdar://95204263

Reviewed by Keith Miller.

This patch adds support for a variadic PAS_ASSERT, and uses it to gather
more debugging info for asserts in pas_local_view_cache_stop and
pas_segregated_heap_ensure_allocator_index.

From manual local disassembly of pas_local_view_cache_stop, I verified that
this patch only changes assertion fail reporting code in an out of line slow
path. The change does perturb code size, and any changes in code may also
cause clang to arbitrarily rearrange the order of some control flow diamonds
(e.g. the cases in a switch statement).

If no extra data is provided to the variadic PAS_ASSERT, it will emit exactly
same code as it does previously when PAS_ASSERT was not variadic. This gives
us confidence that any perturbations in code will only manifest in PAS_ASSERTs
that we choose to add extra info for reporting assertion failures.

Preliminary benchmark results appear to show that performance is neutral.

• Source/bmalloc/libpas/src/libpas/pas_local_view_cache.c:

(pas_local_view_cache_stop):

• Source/bmalloc/libpas/src/libpas/pas_segregated_heap.c:

(pas_segregated_heap_ensure_allocator_index):

• Source/bmalloc/libpas/src/libpas/pas_utils.c:

(pas_crash_with_info_impl1):
(pas_crash_with_info_impl2):
(pas_crash_with_info_impl3):
(pas_crash_with_info_impl4):
(pas_crash_with_info_impl5):
(pas_crash_with_info_impl6):
(pas_report_assertion_failed):

• Source/bmalloc/libpas/src/libpas/pas_utils.h:

(pas_assertion_failed):
(pas_assertion_failed_noreturn_silencer1):
(pas_assertion_failed_noreturn_silencer2):
(pas_assertion_failed_noreturn_silencer3):
(pas_assertion_failed_noreturn_silencer4):
(pas_assertion_failed_noreturn_silencer5):
(pas_assertion_failed_noreturn_silencer6):

Canonical link: ​https://commits.webkit.org/251567@main

8:54 AM Changeset in webkit [295561] by Jonathan Bedard

• 4 edits in trunk/Tools/Scripts/libraries/resultsdbpy

Add git-webkit squash and refactory mock git (Follow-up)
​https://bugs.webkit.org/show_bug.cgi?id=237664
rdar://90040109

Reviewed by Aakash Jain.

• Tools/Scripts/libraries/resultsdbpy/resultsdbpy/init.py: Bump version.
• Tools/Scripts/libraries/resultsdbpy/setup.py: Ditto.
• Tools/Scripts/libraries/resultsdbpy/resultsdbpy/controller/commit_controller_unittest.py:

(CommitControllerTest.test_branches): Add new branch.

• Tools/Scripts/libraries/resultsdbpy/resultsdbpy/model/commit_context_unittest.py:

(CommitContextTest.test_branches): Ditto.

Canonical link: ​https://commits.webkit.org/251566@main

8:39 AM Changeset in webkit [295560] by pvollan@apple.com

• 14 edits
  1 add in trunk/Source

[GPUP] Send Audio component registrations from the UI process
​https://bugs.webkit.org/show_bug.cgi?id=241339
<rdar://88659991>

Reviewed by Geoffrey Garen.

Currently, Audio component registrations are only sent to the WebContent process. They should be sent to the GPU process as well.
This patch moves the existing registration code from the WebContent process code to the Auxiliary process code.

• Source/WebKit/GPUProcess/GPUProcess.messages.in:
• Source/WebKit/Shared/AuxiliaryProcess.h:
• Source/WebKit/Shared/Cocoa/AuxiliaryProcessCocoa.mm:

(WebKit::AuxiliaryProcess::consumeAudioComponentRegistrations):

• Source/WebKit/UIProcess/AuxiliaryProcessProxy.h:
• Source/WebKit/UIProcess/Cocoa/AuxiliaryProcessProcyCocoa.mm: Added.

(WebKit::AuxiliaryProcessProxy::fetchAudioComponentServerRegistrations):

• Source/WebKit/UIProcess/Cocoa/WebProcessProxyCocoa.mm:

(WebKit::WebProcessProxy::sendAudioComponentRegistrations):

• Source/WebKit/UIProcess/GPU/GPUProcessProxy.cpp:

(WebKit::GPUProcessProxy::didFinishLaunching):

• Source/WebKit/WebKit.xcodeproj/project.pbxproj:
• Source/WebKit/WebProcess/WebProcess.h:
• Source/WebKit/WebProcess/cocoa/WebProcessCocoa.mm:

(WebKit::WebProcess::consumeAudioComponentRegistrations): Deleted.

Canonical link: ​https://commits.webkit.org/251565@main

8:34 AM Changeset in webkit [295559] by Razvan Caliman

• 8 edits in trunk

Web Inspector: Remove setting to toggle fuzzy matching for CSS autocompletion
​https://bugs.webkit.org/show_bug.cgi?id=241561

Reviewed by Patrick Angle.

The fuzzy matching CSS autocompletion feature has been enabled by default for a few STP releases already.
No regressions for the autocompletion functionality have been uncovered since.
It's time to remove the setting that allows toggling off the feature.

• LayoutTests/inspector/unit-tests/css-keyword-completions.html:

Turns out the test was never exercising the fuzzy matching branch of autocompletion.
Becaue the useFuzzy flag was missing, it was falling back to testing prefix autocompletion.
The difference is that completion results are not plain strings, but QueryResult instances.

• Source/WebInspectorUI/Localizations/en.lproj/localizedStrings.js:
• Source/WebInspectorUI/UserInterface/Base/Setting.js:
• Source/WebInspectorUI/UserInterface/Models/CSSCompletions.js:

(WI.CSSCompletions.getCompletionText):

Moved the method to get the string value of a completion from WI.CompletionSuggestionsView
to be a static member of WI.CSSCompletions.

• Source/WebInspectorUI/UserInterface/Models/CSSKeywordCompletions.js:

(WI.CSSKeywordCompletions.forPartialPropertyName):

• Source/WebInspectorUI/UserInterface/Views/CompletionSuggestionsView.js:

(WI.CompletionSuggestionsView.prototype.set selectedIndex):
(WI.CompletionSuggestionsView.prototype.getCompletionText): Deleted.

• Source/WebInspectorUI/UserInterface/Views/SettingsTabContentView.js:
• Source/WebInspectorUI/UserInterface/Views/SpreadsheetTextField.js:

(WI.SpreadsheetTextField.prototype._updateCompletions):

Canonical link: ​https://commits.webkit.org/251564@main

7:51 AM Changeset in webkit [295558] by Andres Gonzalez

• 1 edit in trunk/Source/WebCore/accessibility/mac/WebAccessibilityObjectWrapperBase.mm

AX: CrashTracer: com.apple.WebKit.WebContent at com.apple.WebCore: WTF::RetainPtr<NSMutableArray> WTF::createNSArray<WTF::Vector<WTF::RefPtr<WebCore::AXCoreObject, WTF::RawPtrTraits<WebCore::AXCoreObject>, WTF::DefaultRefDerefTraits<WebC. web content
​https://bugs.webkit.org/show_bug.cgi?id=241619

Reviewed by Chris Fleizach.

Check for nullptr before dereferencing the RefPtrs passed to makeNSArray. In some cases, the given AccessibilityChildrenVector may contain nullptrs.

• Source/WebCore/accessibility/mac/WebAccessibilityObjectWrapperBase.mm:

(makeNSArray):

Canonical link: ​https://commits.webkit.org/251563@main

7:02 AM Changeset in webkit [295557] by commit-queue@webkit.org

• 5 edits
  3 adds in trunk/Source

GPUP CGContexts create unattributed Metal resources
​https://bugs.webkit.org/show_bug.cgi?id=241509
rdar://problem/94803367

Patch by Kimmo Kinnunen <​kkinnunen@apple.com> on 2022-06-15
Reviewed by Dean Jackson.

Re-land the patch after revert with disabling the feature for iOS simulator
since task_identity_t is not available in the simulator.

Unattributed Metal related memory allocations grow GPUP footprint,
causing the process go over its memory limit.

Add a CoreGraphics SPI to mark that the CGContext resources belong to a
particular process.

Integrate this via IOSurface::setOwnershipIdentity(), as that is where
the current CGContext for IOSurfaces are instantiated and held.

• Source/WTF/wtf/PlatformHave.h:
• Source/WebCore/PAL/pal/spi/cg/CoreGraphicsSPI.h:
• Source/WebCore/PAL/pal/cg/CoreGraphicsSoftLink.cpp: Added.
• Source/WebCore/PAL/pal/cg/CoreGraphicsSoftLink.h: Added.
• Source/WebCore/PAL/pal/PlatformMac.cmake:
• Source/WebCore/platform/graphics/cocoa/IOSurface.h:
• Source/WebCore/platform/graphics/cocoa/IOSurface.mm:

(WebCore::IOSurface::ensurePlatformContext):
(WebCore::IOSurface::setOwnershipIdentity):

Canonical link: ​https://commits.webkit.org/251562@main

6:43 AM Changeset in webkit [295556] by Alan Bujtas

• 1 edit
  4 adds in trunk

REGRESSION (r289443): Page contents disappear after entering a letter in the username field of bmoharris.com
​https://bugs.webkit.org/show_bug.cgi?id=241625
<rdar://93516876>

Reviewed by Simon Fraser.

overflow: clip forbids scrolling entirely, through any mechanism (​https://drafts.csswg.org/css-overflow/#propdef-overflow)

• LayoutTests/fast/scrolling/scrollIntoView-with-overflow-clip-expected.html: Added.
• LayoutTests/fast/scrolling/scrollIntoView-with-overflow-clip.html: Added.
• LayoutTests/fast/scrolling/selection-reveal-with-overflow-clip-expected.html: Added.
• LayoutTests/fast/scrolling/selection-reveal-with-overflow-clip.html: Added.
• Source/WebCore/rendering/RenderLayer.cpp:

(WebCore::RenderLayer::allowsCurrentScroll const): Make sure the content is not scrollable when overflow: clip is set on the renderer.

Canonical link: ​https://commits.webkit.org/251561@main

4:10 AM Changeset in webkit [295555] by commit-queue@webkit.org

• 9 edits in trunk

[GStreamer][WebRTC] canvas to PeerConnection improvements
​https://bugs.webkit.org/show_bug.cgi?id=241330

Patch by Philippe Normand <​philn@igalia.com> on 2022-06-15
Reviewed by Miguel Gomez and Xabier Rodriguez-Calvar.

Video frames exported from the WebGL canvas need to be flipped on GStreamer ports and their format
should be RGBA. Frames coming from 2D canvas don't need to be flipped and have BGRA format.

Texture flipping is now handled in the player which is useful when painting canvas to a video
element. And flipping is also now handled in RTC outgoing video sources, which is useful when
exporting a canvas to a WebRTC PeerConnection.

The 30fps framerate is hardcoded, ideally should be inferred from the compositor refresh rate.
Having a framerate > 1 is specially needed for the canvas->WebRTC use-case, otherwise a 1/1
framerate would be set in caps, making the corresponding video stream slower than the actual refresh
rate.

• LayoutTests/platform/glib/TestExpectations:
• Source/WebCore/html/HTMLCanvasElement.cpp:

(WebCore::HTMLCanvasElement::toVideoFrame):

• Source/WebCore/platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.cpp:

(WebCore::MediaPlayerPrivateGStreamer::updateTextureMapperFlags):

• Source/WebCore/platform/graphics/gstreamer/VideoFrameGStreamer.cpp:

(WebCore::VideoFrameGStreamer::createFromPixelBuffer):

• Source/WebCore/platform/graphics/gstreamer/VideoFrameGStreamer.h:
• Source/WebCore/platform/graphics/texmap/GraphicsContextGLTextureMapper.cpp:

(WebCore::GraphicsContextGLTextureMapper::paintCompositedResultsToVideoFrame):

• Source/WebCore/platform/mediastream/gstreamer/MockRealtimeVideoSourceGStreamer.cpp:

(WebCore::MockRealtimeVideoSourceGStreamer::updateSampleBuffer):

• Source/WebCore/platform/mediastream/gstreamer/RealtimeOutgoingVideoSourceGStreamer.cpp:

(WebCore::RealtimeOutgoingVideoSourceGStreamer::RealtimeOutgoingVideoSourceGStreamer):
(WebCore::RealtimeOutgoingVideoSourceGStreamer::setPayloadType):

• Source/WebCore/platform/mediastream/gstreamer/RealtimeOutgoingVideoSourceGStreamer.h:

Canonical link: ​https://commits.webkit.org/251560@main

12:55 AM Changeset in webkit [295554] by Adrian Perez de Castro

• 1 edit in trunk/Source/WebInspectorUI/UserInterface/Base/BrowserInspectorFrontendHost.js

[WPE][GTK] REGRESSION(r292084): Browser frontend host is missing recent additions
​https://bugs.webkit.org/show_bug.cgi?id=241594

Reviewed by Patrick Angle.

Update BrowserInspectorFrontendHost.js to add missing methods, update
signatures of changed ones, and remove those which are no longer present
in the IDL interface. This makes the Web Inspector work again when
loaded over HTTP.

• Source/WebInspectorUI/UserInterface/Base/BrowserInspectorFrontendHost.js:

(window.InspectorFrontendHost.WI.BrowserInspectorFrontendHost.prototype.canLoad): Added, always
returns false.
(window.InspectorFrontendHost.WI.BrowserInspectorFrontendHost.prototype.load): Added stub.
(window.InspectorFrontendHost.WI.BrowserInspectorFrontendHost.prototype.getPath): Added stub.
(window.InspectorFrontendHost.WI.BrowserInspectorFrontendHost.prototype.canPickColorFromScreen):
Added, always returns false.
(window.InspectorFrontendHost.WI.BrowserInspectorFrontendHost.prototype.pickColorFromScreen):
Added stub.
(window.InspectorFrontendHost.WI.BrowserInspectorFrontendHost.prototype.getCurrentX): Added stub.
(window.InspectorFrontendHost.WI.BrowserInspectorFrontendHost.prototype.getCurrentY): Added stub.
(window.InspectorFrontendHost.WI.BrowserInspectorFrontendHost.prototype.setPath): Added stub.
(window.InspectorFrontendHost.WI.BrowserInspectorFrontendHost.prototype.get port): Deleted.
(window.InspectorFrontendHost.WI.BrowserInspectorFrontendHost.prototype.append): Deleted.
(window.InspectorFrontendHost.WI.BrowserInspectorFrontendHost.prototype.close): Deleted.

Canonical link: ​https://commits.webkit.org/251559@main