Timeline

May 28, 2021:

10:48 PM Changeset in webkit [278236] by achristensen@apple.com

• 4 edits in trunk

Punycode encode U+0BE6 when not in context of other Tamil characters
​https://bugs.webkit.org/show_bug.cgi?id=226409
<rdar://78160926>

Reviewed by Tim Horton.

Source/WTF:

It has quite legitimate use, so we don't want to always punycode encode it,
but when used in the context of non-Tamil characters we want to punycode encode it.

• wtf/URLHelpers.cpp:

(WTF::URLHelpers::isLookalikeCharacterOfScriptType<USCRIPT_ARMENIAN>):
(WTF::URLHelpers::isLookalikeCharacterOfScriptType<USCRIPT_TAMIL>):
(WTF::URLHelpers::isOfScriptType):
(WTF::URLHelpers::isLookalikeSequence):
(WTF::URLHelpers::isLookalikeCharacter):
(WTF::URLHelpers::isArmenianLookalikeCharacter): Deleted.
(WTF::URLHelpers::isArmenianScriptCharacter): Deleted.
(WTF::URLHelpers::isArmenianLookalikeSequence): Deleted.

Tools:

• TestWebKitAPI/Tests/WTF/cocoa/URLExtras.mm:

(TestWebKitAPI::TEST):

10:27 PM Changeset in webkit [278235] by Wenson Hsieh

• 8 edits
  2 adds in trunk

Web process terminates when attempting to log in on bankid.no
​https://bugs.webkit.org/show_bug.cgi?id=226411
rdar://77419432

Reviewed by Chris Dumez.

Source/WebCore:

On bankid.no, we hit an erroneous MESSAGE_CHECK after decoding and applying an EndClipToDrawingCommands while
the display list replayer's m_maskImageBuffer is null. This may occur as a result of the following sequence of
events:

1. We call into clipToDrawingCommands inside CanvasRenderingContext2DBase, while drawing text with either

a gradient or pattern fill style.

2. A BeginClipToDrawingCommands item is decoded and handled in the GPU process; we create and store a new

mask image buffer on the display list replayer in the GPU process.

3. When we decode the DrawGlyphs item while drawing into the mask buffer, we see that the resource identifier

corresponding to the font is missing from the resource cache in the GPU process, and subsequently go to
sleep while we wait for the resource to arrive.

4. The web font data arrives in the GPU process, and we add it to the cache; this allows us to continue

processing the display list from where we paused replay in step (3).

5. Eventually, we handle the EndClipToDrawingCommands item; however, since this is now happening underneath

a different GPU process wakeup message than in (2), our mask image buffer from (1) is missing. This not only
causes all of the prior drawing commands to target the canvas' image buffer rather than the mask image
buffer, but also causes us to hit a MESSAGE_CHECK and terminate the web process.

To fix this, we add some DisplayList::Replayer::Delegate method hooks to call out into the replayer client when
a mask image buffer has been created or cleared out; in the GPU process, we then use these client hooks to keep
track of the current mask image buffer that is the target of display list replayer.

If we go to sleep and then resume display list processing later for any reason, we then pass this mask image
buffer back to the display list replayer upon construction, and simply let the display list replayer paint into
the maks image buffer's context.

Test: fast/canvas/fill-gradient-text-with-web-font.html

• platform/graphics/displaylists/DisplayListReplayer.cpp:

(WebCore::DisplayList::Replayer::Replayer):
(WebCore::DisplayList::Replayer::applyItem):

Call into the client hooks below.

• platform/graphics/displaylists/DisplayListReplayer.h:

(WebCore::DisplayList::Replayer::Replayer):

When initializing a DisplayList::Replayer, make it possible to take in an initial mask image buffer as well.

(WebCore::DisplayList::Replayer::Delegate::didCreateMaskImageBuffer):
(WebCore::DisplayList::Replayer::Delegate::didResetMaskImageBuffer):

Add new replayer delegate hooks to inform the clients layer when the replayer's mask image buffer has been
created or cleared out. See WebKit ChangeLog for more information.

Source/WebKit:

See WebCore ChangeLog for more details.

• GPUProcess/graphics/RemoteImageBuffer.h:

Implement the new display list replayer delegate hooks and plumb the calls over to the RemoteRenderingBackend.

• GPUProcess/graphics/RemoteRenderingBackend.cpp:

(WebKit::RemoteRenderingBackend::submit):

Keep track of m_currentMaskImageBuffer by implementing the methods below, and use this when initializing our
DisplayList::Replayer if needed.

(WebKit::RemoteRenderingBackend::didCreateMaskImageBuffer):
(WebKit::RemoteRenderingBackend::didResetMaskImageBuffer):

• GPUProcess/graphics/RemoteRenderingBackend.h:

LayoutTests:

Add a layout test that exercises the crash by first serializing a large number of display list items, and then
calling fillText with a web font and a gradient fillStyle.

• fast/canvas/fill-gradient-text-with-web-font-expected.html: Added.
• fast/canvas/fill-gradient-text-with-web-font.html: Added.

9:34 PM Changeset in webkit [278234] by zhifei_fang@apple.com

• 2 edits in trunk/Tools

[JSC] Escape $ only when run with SSH
​https://bugs.webkit.org/show_bug.cgi?id=226385

Reviewed by Alexey Proskuryakov.

• Scripts/run-jsc-stress-tests:

6:28 PM Changeset in webkit [278233] by Chris Dumez

• 5 edits in trunk/Source/WebCore

Regression: Raw AudioBufferSourceNode playback causes repeated crackling sound
​https://bugs.webkit.org/show_bug.cgi?id=222098
<rdar://problem/74546471>

Reviewed by Darin Adler.

The issue was due to certain audio nodes (such as AudioBufferSourceNode) staying in the audio
graph after they were no longer needed. As a result, the audio graph was becoming larger and
larger and the audio performance should degrade due to us traversing this increasingly large
graph.

The audio nodes that had trouble getting removed from the graphs were tail processing nodes.
Those are nodes that may still produce output for a while after they no longer have any
inputs. PannerNode and DelayNode are example of such nodes.

When a Node is no longer useful (the node no longer has any connections, m_connectionRefCount
is 0), we call disableOutputsIfNecessary() on it to disable its outputs and avoid doing any
processing of these outputs since they no longer have an input. disableOutputsIfNecessary()
would normally call disableOutputs() (if not already disabled) but there was an exception
in the case where the node requires tail processing. For such nodes, you wouldn't want to
disable them until they've finished processing their tail.

The issue was that once those nodes had finished processing their tail, we wouldn't come
back to them and disable their outputs later on. As a result, we would process more and
more (silent) nodes and the audio performance would seriously deteriorate.

To address this, we now add the node to the context's list of tail processing nodes in
disableOutputsIfNecessary() if the node requires tail processing. After each rendering
quantum, we go through those tail processing nodes and check if they have finished
processing their tail. If they have, we go ahead and disable their outputs at this point
and remove them from the list.

This is modeled after what Blink is doing for tail processing nodes in:

• ​https://github.com/chromium/chromium/blob/master/third_party/blink/renderer/modules/webaudio/deferred_task_handler.cc

I have verified that on the following demo, the performance no longer deteriorates after
a while:

• ​https://jsfiddle.net/KrisJohnson/s5vL24o1/123/

• Modules/webaudio/AudioNode.cpp:

(WebCore::AudioNode::enableOutputsIfNecessary):
(WebCore::AudioNode::disableOutputsIfNecessary):

• Modules/webaudio/AudioNode.h:

(WebCore::AudioNode::isTailProcessing const):
(WebCore::AudioNode::setIsTailProcessing):

• Modules/webaudio/BaseAudioContext.cpp:

(WebCore::BaseAudioContext::uninitialize):
(WebCore::BaseAudioContext::handlePostRenderTasks):
(WebCore::BaseAudioContext::addTailProcessingNode):
(WebCore::BaseAudioContext::removeTailProcessingNode):
(WebCore::BaseAudioContext::updateTailProcessingNodes):
(WebCore::BaseAudioContext::disableOutputsForFinishedTailProcessingNodes):
(WebCore::BaseAudioContext::finishTailProcessing):

• Modules/webaudio/BaseAudioContext.h:

(WebCore::BaseAudioContext::TailProcessingNode::TailProcessingNode):
(WebCore::BaseAudioContext::TailProcessingNode::~TailProcessingNode):
(WebCore::BaseAudioContext::TailProcessingNode::operator-> const):
(WebCore::BaseAudioContext::TailProcessingNode::operator== const):

6:05 PM Changeset in webkit [278232] by Chris Dumez

• 9 edits in trunk

Stop using UncheckedLock in WTF::MetaAllocator
​https://bugs.webkit.org/show_bug.cgi?id=226396

Reviewed by Darin Adler.

Stop using UncheckedLock in WTF::MetaAllocator, as it is being phased out in favor on Lock,
which supports Clang thread safety analysis.

Source/JavaScriptCore:

• jit/ExecutableAllocator.cpp:

(JSC::ExecutableAllocator::getLock const):

• jit/ExecutableAllocator.h:

(JSC::ExecutableAllocatorBase::WTF_RETURNS_LOCK):

• tools/VMInspector.cpp:

(JSC::VMInspector::isValidExecutableMemory):

Source/WTF:

• wtf/MetaAllocator.cpp:

(WTF::MetaAllocator::release):
(WTF::MetaAllocator::MetaAllocator):
(WTF::MetaAllocator::allocate):
(WTF::MetaAllocator::currentStatistics):

• wtf/MetaAllocator.h:

5:52 PM Changeset in webkit [278231] by Chris Dumez

• 6 edits in trunk/Source/WebCore

DelayDSPKernel::process() is slow
​https://bugs.webkit.org/show_bug.cgi?id=226358

Reviewed by Darin Adler.

When I profiled the demo at ​https://jsfiddle.net/KrisJohnson/s5vL24o1/123/ (in the context of Bug 222098),
I noticed that 20% of the CPU time was spent under DelayDSPKernel::process().

To improve this, we now vectorize DelayDSPKernel::process() in the common case where there is no automation
and the delay time is constant.

The implementation is very similar to the one in Blink:

• ​https://github.com/chromium/chromium/blob/master/third_party/blink/renderer/platform/audio/audio_delay_dsp_kernel.cc

Some differences compared to the Blink implementation:

• I did not vectorize the A-rate case for simplicity. It is not as common and it is more complicated. We may consider doing this in the future if really needed.
• On Cocoa, we leveage Accelerate's vDSP_vintb() to do the interpolation instead of doing 2 separate operations.

This doesn't fix Bug 222098 but it does improve the situation quite a bit. I also see that the CPU time
spent under DelayDSPKernel::process() went from ~20% to 1.2% on this demo.

No new tests, no Web-facing behavior change, just a performance optimization.

• Modules/webaudio/DelayDSPKernel.cpp:

(WebCore::copyToCircularBuffer):
(WebCore::DelayDSPKernel::DelayDSPKernel):
(WebCore::DelayDSPKernel::bufferLengthForDelay const):
(WebCore::DelayDSPKernel::process):
(WebCore::DelayDSPKernel::processARate):
(WebCore::DelayDSPKernel::processKRate):

• Modules/webaudio/DelayDSPKernel.h:
• platform/audio/VectorMath.cpp:

(WebCore::VectorMath::substract):
(WebCore::VectorMath::interpolate):

• platform/audio/VectorMath.h:

5:18 PM Changeset in webkit [278230] by Brent Fulgham

• 2 edits in trunk/Source/WebCore

[Cocoa] Prevent GPU Process from attempt to connect to the AppSSO service (Part 2)
​https://bugs.webkit.org/show_bug.cgi?id=226393
<rdar://problem/78614833>

Reviewed by Eric Carlson.

In Bug 224834 I modified WebKit to use avoid invoking AppSSO in CFNetwork calls, since we make those
decisions prior to beginning our network interactions. I missed a media call where this was being done.

This patch corrects that mistake.

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

(WebCore::canonicalURL):

4:53 PM Changeset in webkit [278229] by Chris Dumez

• 11 edits
  1 delete in trunk/Source

Stop using UncheckedLock in JSDOMGlobalObject
​https://bugs.webkit.org/show_bug.cgi?id=226281

Reviewed by Darin Adler.

Source/JavaScriptCore:

Drop JSC::lockDuringMarking() and its now empty header given that it does conditional
locking and it is not compatible with Clang thread safety analysis.

• CMakeLists.txt:
• JavaScriptCore.xcodeproj/project.pbxproj:
• bytecode/UnlinkedCodeBlock.h:
• heap/LockDuringMarking.h: Removed.

Source/WebCore:

Stop using UncheckedLock in JSDOMGlobalObject and use Lock instead of benefit from Clang thread
safety analysis. This is a step towards phasing out UncheckedLock and improving the thread
safety of our code.

• bindings/js/JSDOMGlobalObject.cpp:

(WebCore::JSDOMGlobalObject::visitChildrenImpl):
(WebCore::JSDOMGlobalObject::clearDOMGuardedObjects):

• bindings/js/JSDOMGlobalObject.h:

(WebCore::getDOMConstructor):

• bindings/js/JSDOMGuardedObject.cpp:

(WebCore::DOMGuardedObject::DOMGuardedObject):
(WebCore::DOMGuardedObject::clear):
(WebCore::DOMGuardedObject::removeFromGlobalObject):

• bindings/js/JSDOMGuardedObject.h:
• bindings/js/JSDOMWrapperCache.cpp:

(WebCore::cacheDOMStructure):

3:52 PM Changeset in webkit [278228] by Russell Epstein

• 1 copy in tags/Safari-612.1.15.0.9

Tag Safari-612.1.15.0.9.

2:53 PM Changeset in webkit [278227] by don.olmstead@sony.com

• 4 edits in trunk/Source/WebKit

[Cocoa] HDR variants in HLS playlists not selected when screen is HDR
​https://bugs.webkit.org/show_bug.cgi?id=226349
<rdar://76963977>

Unreviewed build fix after r278204.

Update the method signature of RemoteMediaPlayerProxy::prepareForPlayback for non-Cocoa
platforms.

• GPUProcess/media/gstreamer/RemoteMediaPlayerProxyGStreamer.cpp:

(WebKit::RemoteMediaPlayerProxy::prepareForPlayback):

• GPUProcess/media/playstation/RemoteMediaPlayerProxyPlayStation.cpp:

(WebKit::RemoteMediaPlayerProxy::prepareForPlayback):

• GPUProcess/media/win/RemoteMediaPlayerProxyWin.cpp:

(WebKit::RemoteMediaPlayerProxy::prepareForPlayback):

2:53 PM Changeset in webkit [278226] by sihui_liu@apple.com

• 2 edits in trunk/Source/WebKit

Regression(r278179): m_serverLock is still held after WebIDBServer is destroyed
​https://bugs.webkit.org/show_bug.cgi?id=226389

Reviewed by Chris Dumez.

• NetworkProcess/IndexedDB/WebIDBServer.cpp:

(WebKit::WebIDBServer::close):

1:18 PM Changeset in webkit [278225] by commit-queue@webkit.org

• 5 edits in trunk

Prevent resetting scroll position in unnecessary cases during page load
​https://bugs.webkit.org/show_bug.cgi?id=226338
rdar://75334651

Patch by Matt Gilligan <​matthew_gilligan@apple.com> on 2021-05-28
Reviewed by Simon Fraser.

Source/WebKit:

When WKWebView's scroll view has a refresh control and is rubber banding along the top
edge, don't cancel scrolling animations or reset contentOffset within valid range due to
the page load. This can cause flickering during dragging and can prevent the bounce
animation from finishing.

Test: ScrollViewInsetTests.PreserveContentOffsetForRefreshControl

• UIProcess/API/ios/WKWebViewIOS.mm:

(-[WKWebView _didCommitLoadForMainFrame]):
Avoid calling -_stopScrollingAndZoomingAnimations when rubber banding for a refresh control
to prevent cancelling a bounce animation.

(-[WKWebView _restoreScrollAndZoomStateForTransaction:]):
(-[WKWebView _didCommitLayerTree:]):
Don't adjust content offset to be within valid range when rubber banding for a refresh
control, or the scroll position will be reset while dragging or bouncing.

(-[WKWebView _scrollViewIsRubberBandingForRefreshControl]):
Return whether _scrollView has a refresh control and is rubber banding along the top edge.

• UIProcess/ios/WKScrollView.mm:

(-[WKScrollView _setContentSizePreservingContentOffsetDuringRubberband:]):
When the refresh control is dismissing, the scroll view will be rubber banding but not
dragging or bouncing. Treat a non-nil refreshControl as a hint that the scroll view might
be rubber banding.

Tools:

• TestWebKitAPI/Tests/ios/ScrollViewInsetTests.mm:

(TestWebKitAPI::TEST):

1:17 PM Changeset in webkit [278224] by rmorisset@apple.com

• 3 edits
  1 add in trunk

Fix LikelyDenseUnsignedIntegerSet::clear()
​https://bugs.webkit.org/show_bug.cgi?id=226388
JSTests:

rdar://78607433

Reviewed by Mark Lam.

• stress/stack-allocation-regression.js: Added.

(foo):

Source/WTF:

Reviewed by Mark Lam.

There are two problems with it:
1) It calls BitVector::clearAll(), which does not free any memory.
Instead, it should call BitVector::~BitVector(), then do a placement new of a fresh BitVector (to get it back to its inline condition)
2) More problematically, it changes m_size before calling isBitVector() which relies crucially on the value of m_size.
So it is going to believe that it is in BitVector mode even when it is actually in HashSet mode.

• wtf/LikelyDenseUnsignedIntegerSet.h:

(WTF::LikelyDenseUnsignedIntegerSet::clear):

1:13 PM Changeset in webkit [278223] by jer.noble@apple.com

• 3 edits in trunk/Source/WebKit

[Cocoa] Hang in WebM format reader when loading content from ​https://powerrangers.fandom.com/
​https://bugs.webkit.org/show_bug.cgi?id=226128
<rdar://77645877>

Reviewed by Eric Carlson.

When creating a format reader, MediaToolbox will query for various properties including ones about
MetadataFormat and MetadataKeySpace. We don't support those properties, but we still block until
tracks are loaded before answering that we don't support them. Instead, just return early unless
the query is for the one property which we do support, Duration.

• Shared/mac/MediaFormatReader/MediaFormatReader.cpp:

(WebKit::MediaFormatReader::copyProperty):

1:08 PM Changeset in webkit [278222] by Jean-Yves Avenard

• 5 edits
  2 adds in trunk

Add MediaSession.callActionHandler
​https://bugs.webkit.org/show_bug.cgi?id=226077
rdar://77463304

Reviewed by Eric Carlson.

Source/WebCore:

Make it possible for JavaScript to be able to trigger a MediaSession action
handler so a MediaSession.coordinator polyfill such as one provided by an
external JS library will work.
Test: media/media-session/callActionHandler.html

• Modules/mediasession/MediaSession.cpp:

(WebCore::MediaSession::callActionHandler):

• Modules/mediasession/MediaSession.h:
• Modules/mediasession/MediaSession.idl: Add method definition.

LayoutTests:

• media/media-session/callActionHandler-expected.txt: Added.
• media/media-session/callActionHandler.html: Added.

1:02 PM Changeset in webkit [278221] by Wenson Hsieh

• 2 edits in trunk/Source/WebKit

[iOS] Adjust logic when determining whether to enable image extraction actions in the callout bar
​https://bugs.webkit.org/show_bug.cgi?id=226387
rdar://76781788

Reviewed by Tim Horton and Devin Rousso.

Call into the superclass (which contains additional restrictions beyond our isContentEditable check on the
editor state), and additionally refactor this helper function to take a selector and return whether or not it is
one of the image extraction action selectors (as opposed to returning the selector itself).

See Radar for more details.

• UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView canPerformActionForWebView:withSender:]):

12:54 PM Changeset in webkit [278220] by Alan Coon

• 8 edits in branches/safari-612.1.15.0-branch/Source

Versioning.

WebKit-7612.1.15.0.9

12:51 PM Changeset in webkit [278219] by Alan Bujtas

• 5 edits
  2 adds in trunk

CrashOnOverflow in WebCore::RenderTable::cellBefore(WebCore::RenderTableCell const*)
​https://bugs.webkit.org/show_bug.cgi?id=225949

Reviewed by Antti Koivisto.

Source/WebCore:

The table section maintains a grid of rows and columns. This grid is used to jump to cells directly during layout (as opposed to tree walking).
It needs to be re-populated on tree mutations e.g when a new row is added to the table.
RenderTableSection::willInsertTableRow assumes that rows and cells are inserted subsequently, after each other and
it's ok to ignore the grid invalidation on appends as the subsequent ::addCell() call takes care of the grid.
However in some continuation split cases, we re-parent complete rows under table sections which means no subsequent ::addCell() calls.

This patch addresses this case by using the same pattern we have for block continuation invalidation in RenderTreeBuilder::splitAnonymousBoxesAroundChild().

Test: fast/table/table-split-on-insert-with-section-crash.html

• rendering/updating/RenderTreeBuilder.cpp:

(WebCore::RenderTreeBuilder::markBoxForRelayoutAfterSplit):
(WebCore::markBoxForRelayoutAfterSplit): Deleted.

• rendering/updating/RenderTreeBuilder.h:
• rendering/updating/RenderTreeBuilderInline.cpp:

(WebCore::RenderTreeBuilder::Inline::splitInlines):

LayoutTests:

• fast/table/table-split-on-insert-with-section-crash-expected.txt: Added.
• fast/table/table-split-on-insert-with-section-crash.html: Added.

12:40 PM Changeset in webkit [278218] by zhifei_fang@apple.com

• 2 edits in trunk/Tools

[JSC] Fix run-jsc-stress-tests missing escape $
​https://bugs.webkit.org/show_bug.cgi?id=226385

Reviewed by Alexey Proskuryakov.

• Scripts/run-jsc-stress-tests:

11:54 AM Changeset in webkit [278217] by weinig@apple.com

• 17 edits in trunk/Source

Add stub implementation of CA separated portal bits for GraphicsLayer
​https://bugs.webkit.org/show_bug.cgi?id=226343

Reviewed by Simon Fraser.

Source/WebCore:

Plumbs isSeparatedPortal and isDescendentOfSeparatedPortal bits through GraphicsLayer.
Renames setSeparated to setIsSeparated for consistency.

Currently unused, but getting the bits in is big enough that its helpful
to land on its own.

• platform/graphics/GraphicsLayer.cpp:

(WebCore::GraphicsLayer::GraphicsLayer):

• platform/graphics/GraphicsLayer.h:

(WebCore::GraphicsLayer::isIsSeparated const):
(WebCore::GraphicsLayer::setIsSeparated):
(WebCore::GraphicsLayer::isSeparatedPortal const):
(WebCore::GraphicsLayer::setIsSeparatedPortal):
(WebCore::GraphicsLayer::isDescendentOfSeparatedPortal const):
(WebCore::GraphicsLayer::setIsDescendentOfSeparatedPortal):
(WebCore::GraphicsLayer::isSeparated const): Deleted.
(WebCore::GraphicsLayer::setSeparated): Deleted.

• platform/graphics/ca/GraphicsLayerCA.cpp:

(WebCore::GraphicsLayerCA::setIsSeparated):
(WebCore::GraphicsLayerCA::setIsSeparatedPortal):
(WebCore::GraphicsLayerCA::setIsDescendentOfSeparatedPortal):
(WebCore::GraphicsLayerCA::commitLayerChangesBeforeSublayers):
(WebCore::GraphicsLayerCA::updateIsSeparated):
(WebCore::GraphicsLayerCA::updateIsSeparatedPortal):
(WebCore::GraphicsLayerCA::updateIsDescendentOfSeparatedPortal):
(WebCore::GraphicsLayerCA::setSeparated): Deleted.
(WebCore::GraphicsLayerCA::updateSeparated): Deleted.

• platform/graphics/ca/GraphicsLayerCA.h:
• platform/graphics/ca/PlatformCALayer.h:
• platform/graphics/ca/cocoa/PlatformCALayerCocoa.h:
• platform/graphics/ca/cocoa/PlatformCALayerCocoa.mm:

(WebCore::PlatformCALayerCocoa::setIsSeparated):
(WebCore::PlatformCALayerCocoa::isSeparatedPortal const):
(WebCore::PlatformCALayerCocoa::setIsSeparatedPortal):
(WebCore::PlatformCALayerCocoa::isDescendentOfSeparatedPortal const):
(WebCore::PlatformCALayerCocoa::setIsDescendentOfSeparatedPortal):
(WebCore::PlatformCALayerCocoa::setSeparated): Deleted.

• rendering/RenderLayerBacking.cpp:

(WebCore::RenderLayerBacking::updateGeometry):

Source/WebKit:

Plumbs isSeparatedPortal and isDescendentOfSeparatedPortal bits through GraphicsLayer.
Renames setSeparated to setIsSeparated for consistency.

Currently unused, but getting the bits in is big enough that its helpful
to land on its own.

• Shared/RemoteLayerTree/RemoteLayerTreePropertyApplier.mm:

(WebKit::RemoteLayerTreePropertyApplier::applyPropertiesToLayer):

• Shared/RemoteLayerTree/RemoteLayerTreeTransaction.h:
• Shared/RemoteLayerTree/RemoteLayerTreeTransaction.mm:

(WebKit::RemoteLayerTreeTransaction::LayerProperties::LayerProperties):
(WebKit::RemoteLayerTreeTransaction::LayerProperties::encode const):
(WebKit::RemoteLayerTreeTransaction::LayerProperties::decode):
(WebKit::dumpChangedLayers):
(WebKit::RemoteLayerTreeTransaction::RemoteLayerTreeTransaction): Deleted.
(WebKit::RemoteLayerTreeTransaction::~RemoteLayerTreeTransaction): Deleted.

• WebProcess/WebPage/RemoteLayerTree/PlatformCALayerRemote.cpp:

(WebKit::PlatformCALayerRemote::setIsSeparated):
(WebKit::PlatformCALayerRemote::isSeparatedPortal const):
(WebKit::PlatformCALayerRemote::setIsSeparatedPortal):
(WebKit::PlatformCALayerRemote::isDescendentOfSeparatedPortal const):
(WebKit::PlatformCALayerRemote::setIsDescendentOfSeparatedPortal):
(WebKit::PlatformCALayerRemote::setSeparated): Deleted.

• WebProcess/WebPage/RemoteLayerTree/PlatformCALayerRemote.h:

Source/WTF:

• wtf/PlatformHave.h: Add HAVE_CORE_ANIMATION_SEPARATED_PORTALS.

11:47 AM Changeset in webkit [278216] by Russell Epstein

• 1 copy in tags/Safari-611.3.6.1.3

Tag Safari-611.3.6.1.3.

11:45 AM Changeset in webkit [278215] by Russell Epstein

• 2 edits in branches/safari-611.3.6.1-branch/Source/WTF

Cherry-pick r278209. rdar://problem/78595584

Unreviewed build fix after r277881.

Deal with IOPOL_TYPE_VFS_MATERIALIZE_DATALESS_FILES potentially not being defined on
older OSes.

• wtf/cocoa/FileSystemCocoa.mm: (WTF::FileSystemImpl::setAllowsMaterializingDatalessFiles): (WTF::FileSystemImpl::allowsMaterializingDatalessFiles):

git-svn-id: ​https://svn.webkit.org/repository/webkit/trunk@278209 268f45cc-cd09-0410-ab3c-d52691b4dbfc

11:45 AM Changeset in webkit [278214] by Russell Epstein

• 2 edits in branches/safari-611.3.6.1-branch/Source/WTF

Cherry-pick r278208. rdar://problem/78595584

Unreviewed build fix after r277881.

Add missing header include.

• wtf/cocoa/FileSystemCocoa.mm:

git-svn-id: ​https://svn.webkit.org/repository/webkit/trunk@278208 268f45cc-cd09-0410-ab3c-d52691b4dbfc

11:36 AM Changeset in webkit [278213] by sbarati@apple.com

• 11 edits
  2 adds in trunk/Source/JavaScriptCore

Add the ability to dump statistics about cumulative counts and code sizes of Baseline JIT opcodes and DFG nodes
​https://bugs.webkit.org/show_bug.cgi?id=226200

Reviewed by Robin Morisset.

• JavaScriptCore.xcodeproj/project.pbxproj:
• Sources.txt:
• dfg/DFGSlowPathGenerator.h:

(JSC::DFG::SlowPathGenerator::currentNode const):

• dfg/DFGSpeculativeJIT.cpp:

(JSC::DFG::SpeculativeJIT::runSlowPathGenerators):
(JSC::DFG::SpeculativeJIT::compileCurrentBlock):

• jit/JIT.cpp:

(JSC::JIT::privateCompileMainPass):
(JSC::JIT::privateCompileSlowCases):
(JSC::JIT::compileAndLinkWithoutFinalizing):

• jit/JITSizeStatistics.cpp: Added.

(JSC::JITSizeStatistics::markStart):
(JSC::JITSizeStatistics::markEnd):
(JSC::JITSizeStatistics::dump const):

• jit/JITSizeStatistics.h: Added.

(JSC::JITSizeStatistics::reset):

• jsc.cpp:

(runJSC):

• runtime/OptionsList.h:
• runtime/VM.cpp:

(JSC::VM::VM):

• runtime/VM.h:
• tools/JSDollarVM.cpp:

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

11:31 AM Changeset in webkit [278212] by Russell Epstein

• 8 edits in branches/safari-611.3.6.1-branch/Source

Versioning.

WebKit-7611.3.6.1.3

11:13 AM Changeset in webkit [278211] by sbarati@apple.com

• 3 edits
  1 add in trunk

Don't sink arguments past the context of the inline call frame they were created in
​https://bugs.webkit.org/show_bug.cgi?id=226363
<rdar://78392801>

Reviewed by Filip Pizlo.

JSTests:

• stress/dont-sink-arguments-past-inline-call-frame.js: Added.

(foo):
(fooWrap):
(empty):
(bar):

Source/JavaScriptCore:

DFG OSR exit isn't set up to allow materializing phantom arguments objects
when the inline call frame is no longer present in the exit metadata. There
may be multiple reasons for this, but there's at least one I discovered:
We don't analyze any locals when reconstructing the variable event stream
if that local is past the extent of the inline call frame.

This patch makes it so we treat it as an escape if between the definition and
the last use of a sink candidate (either via an IR use or bytecode liveness)
there is an exit to a place in the graph where the inline call frame of the
definition is no longer present.

• dfg/DFGVarargsForwardingPhase.cpp:

11:11 AM Changeset in webkit [278210] by stephan.szabo@sony.com

• 14 edits in trunk/Source/JavaScriptCore

Regenerate builtins generator expectations after optional changes
​https://bugs.webkit.org/show_bug.cgi?id=226381

Reviewed by Chris Dumez.

• Scripts/tests/builtins/expected/JavaScriptCore-Builtin.Promise-Combined.js-result:
• Scripts/tests/builtins/expected/JavaScriptCore-Builtin.Promise-Separate.js-result:
• Scripts/tests/builtins/expected/JavaScriptCore-Builtin.prototype-Combined.js-result:
• Scripts/tests/builtins/expected/JavaScriptCore-Builtin.prototype-Separate.js-result:
• Scripts/tests/builtins/expected/JavaScriptCore-BuiltinConstructor-Combined.js-result:
• Scripts/tests/builtins/expected/JavaScriptCore-BuiltinConstructor-Separate.js-result:
• Scripts/tests/builtins/expected/JavaScriptCore-InternalClashingNames-Combined.js-result:
• Scripts/tests/builtins/expected/WebCore-AnotherGuardedInternalBuiltin-Separate.js-result:
• Scripts/tests/builtins/expected/WebCore-ArbitraryConditionalGuard-Separate.js-result:
• Scripts/tests/builtins/expected/WebCore-GuardedBuiltin-Separate.js-result:
• Scripts/tests/builtins/expected/WebCore-GuardedInternalBuiltin-Separate.js-result:
• Scripts/tests/builtins/expected/WebCore-UnguardedBuiltin-Separate.js-result:
• Scripts/tests/builtins/expected/WebCore-xmlCasingTest-Separate.js-result:

10:59 AM Changeset in webkit [278209] by Chris Dumez

• 2 edits in trunk/Source/WTF

Unreviewed build fix after r277881.

Deal with IOPOL_TYPE_VFS_MATERIALIZE_DATALESS_FILES potentially not being defined on
older OSes.

• wtf/cocoa/FileSystemCocoa.mm:

(WTF::FileSystemImpl::setAllowsMaterializingDatalessFiles):
(WTF::FileSystemImpl::allowsMaterializingDatalessFiles):

10:47 AM Changeset in webkit [278208] by Chris Dumez

• 2 edits in trunk/Source/WTF

Unreviewed build fix after r277881.

Add missing header include.

• wtf/cocoa/FileSystemCocoa.mm:

10:23 AM Changeset in webkit [278207] by Brent Fulgham

• 2 edits in trunk/Source/WebCore

Buffer computations should use correct type
​https://bugs.webkit.org/show_bug.cgi?id=226352
<rdar://problem/78116213>

Reviewed by Ryosuke Niwa.

The implementation of utf8Buffer holds the buffer size in an 'int', even though our buffer
APIs take size_t arguments. Let's make this consistent.

• platform/SharedBuffer.cpp:

(WebCore::utf8Buffer):

9:58 AM Changeset in webkit [278206] by mark.lam@apple.com

• 2 edits in trunk/Source/JavaScriptCore

The jsc shell should not continue executing after a termination has been received.
​https://bugs.webkit.org/show_bug.cgi?id=226368
rdar://78013374

Reviewed by Saam Barati.

This patch makes the jsc shell forbid execution in all the places that it has
caught and seen a TerminationException.

• jsc.cpp:

(JSC_DEFINE_HOST_FUNCTION):
(checkException):
(runWithOptions):
(runInteractive):

9:52 AM Changeset in webkit [278205] by Amir Mark Jr.

• 2 edits in trunk/LayoutTests

[BigSur Wk1] imported/w3c/web-platform-tests/css/css-contain/contain-paint-ignored-cases-ruby-stacking-and-clipping-001.html is a flaky failure
​https://bugs.webkit.org/show_bug.cgi?id=226377

Unreviewed test gardening.

• platform/mac-wk1/TestExpectations:

9:44 AM Changeset in webkit [278204] by jer.noble@apple.com

• 8 edits in trunk/Source/WebKit

[Cocoa] HDR variants in HLS playlists not selected when screen is HDR
​https://bugs.webkit.org/show_bug.cgi?id=226349
<rdar://76963977>

Reviewed by Eric Carlson.

An existing test, http/tests/media/hls/hls-hdr-switch.html, should have caught this regression,
but is not currently functional as it depends on PHP, which has been removed.

The preferredDynamicRangeMode needs to be piped over from WebContent -> GPU process.

• GPUProcess/media/RemoteMediaPlayerProxy.cpp:

(WebKit::RemoteMediaPlayerProxy::setPreferredDynamicRangeMode):

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

(WebKit::RemoteMediaPlayerProxy::prepareForPlayback):

• Scripts/webkit/messages.py:

(headers_for_type):

• WebProcess/GPU/media/MediaPlayerPrivateRemote.cpp:

(WebKit::MediaPlayerPrivateRemote::prepareForPlayback):
(WebKit::MediaPlayerPrivateRemote::setPreferredDynamicRangeMode):

• WebProcess/GPU/media/MediaPlayerPrivateRemote.h:

9:43 AM Changeset in webkit [278203] by eric.carlson@apple.com

• 2 edits in trunk/Source/WebCore

[Cocoa] Return immediately when asked to paint an AVPlayer that hasn't reached HaveCurrentData
​https://bugs.webkit.org/show_bug.cgi?id=226350
rdar://77910715

Reviewed by Jer Noble.

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

(WebCore::MediaPlayerPrivateAVFoundationObjC::updateLastPixelBuffer): Return immediately
if readyState < HaveCurrentData.
(WebCore::MediaPlayerPrivateAVFoundationObjC::videoOutputHasAvailableFrame): Ditto.
(WebCore::MediaPlayerPrivateAVFoundationObjC::updateLastImage): Ditto.

9:39 AM Changeset in webkit [278202] by Devin Rousso

• 8 edits in trunk

[Modern Media Controls] put the localized information for tracks first if the track is not in the user's preferred language
​https://bugs.webkit.org/show_bug.cgi?id=226361
<rdar://problem/78497853>

Reviewed by Eric Carlson.

Source/WebCore:

Test: media/modern-media-controls/tracks-support/sorted-by-user-preferred-languages.html

This makes it clearer/easier for the user to understand what's going on when the track label
is not in the user's preferred language. For example, if a <video> has an "English" audio
track, it's not very helpful to show "English (英语)" for Chinese users instead of something
like "英语 (English)".

If the track has a label and either matches the user's preferred language or its label
already contains the localized language name, combine the track's label with a localized
description of the track if that description isn't already in the label. Otherwise, add
the localized language name first, then the description of the track, and finally the label.

Here are some examples if the user's preferred language is English:

• <track kind="subtitles" srclang="en"> would show "English"
• <track kind="subtitles" srclang="en" label="Foo"> would show "Foo"
• <track kind="captions" srclang="en"> would show "English Captions"
• <track kind="captions" srclang="en" label="Foo"> would show "Foo Captions"
• <track kind="captions" srclang="en" label="Captions"> would show "Captions"
• <track kind="subtitles" srclang="fr"> would show "French"
• <track kind="subtitles" srclang="fr" label="Foo"> would show "French (Foo)"
• <track kind="captions" srclang="fr"> would show "French Captions"
• <track kind="captions" srclang="fr" label="Foo"> would show "French Captions (Foo)"
• <track kind="captions" srclang="fr" label="Captions"> would show "French Captions"

• page/CaptionUserPreferences.cpp:

(WebCore::trackDisplayName):

• page/CaptionUserPreferencesMediaAF.cpp:

(WebCore::addTextTrackKindDisplayNameIfNeeded): Added.
(WebCore::addAudioTrackKindDisplayNameIfNeeded): Added.
(WebCore::addTrackKindDisplayNameIfNeeded): Added.
(WebCore::trackDisplayName):
(WebCore::CaptionUserPreferencesMediaAF::displayNameForTrack const):
(WebCore::textTrackCompare):
(WebCore::CaptionUserPreferencesMediaAF::sortedTrackListForMenu):
(WebCore::buildDisplayStringForTrackBase): Deleted.

• platform/LocalizedStrings.h:
• platform/LocalizedStrings.cpp:

(WebCore::trackNoLabelText): Added.
(WebCore::textTrackOffMenuItemText): Added.
(WebCore::textTrackAutomaticMenuItemText): Added.
(WebCore::addTrackLabelAsSuffix): Added.
(WebCore::textTrackKindClosedCaptionsDisplayName): Added.
(WebCore::addTextTrackKindClosedCaptionsSuffix): Added.
(WebCore::textTrackKindCaptionsDisplayName): Added.
(WebCore::addTextTrackKindCaptionsSuffix): Added.
(WebCore::textTrackKindDescriptionsDisplayName): Added.
(WebCore::addTextTrackKindDescriptionsSuffix): Added.
(WebCore::textTrackKindChaptersDisplayName): Added.
(WebCore::addTextTrackKindChaptersSuffix): Added.
(WebCore::textTrackKindMetadataDisplayName): Added.
(WebCore::addTextTrackKindMetadataSuffix): Added.
(WebCore::textTrackKindSDHDisplayName): Added.
(WebCore::addTextTrackKindSDHSuffix): Added.
(WebCore::textTrackKindEasyReaderDisplayName): Added.
(WebCore::addTextTrackKindEasyReaderSuffix): Added.
(WebCore::textTrackKindForcedDisplayName): Added.
(WebCore::addTextTrackKindForcedSuffix): Added.
(WebCore::audioTrackKindDescriptionsDisplayName): Added.
(WebCore::addAudioTrackKindDescriptionsSuffix): Added.
(WebCore::audioTrackKindCommentaryDisplayName): Added.
(WebCore::addAudioTrackKindCommentarySuffix): Added.
(WebCore::textTrackNoLabelText): Deleted.
(WebCore::audioTrackNoLabelText): Deleted.
(WebCore::captionsTextTrackKindDisplayName): Deleted.
(WebCore::captionsTextTrackWithoutLabelMenuItemText): Deleted.
(WebCore::descriptionsTextTrackKindDisplayName): Deleted.
(WebCore::descriptionsTextTrackWithoutLabelMenuItemText): Deleted.
(WebCore::chaptersTextTrackKindDisplayName): Deleted.
(WebCore::chaptersTextTrackWithoutLabelMenuItemText): Deleted.
(WebCore::metadataTextTrackKindDisplayName): Deleted.
(WebCore::metadataTextTrackWithoutLabelMenuItemText): Deleted.
(WebCore::textTrackCountryAndLanguageMenuItemText): Deleted.
(WebCore::textTrackLanguageMenuItemText): Deleted.
(WebCore::closedCaptionKindTrackDisplayName): Deleted.
(WebCore::closedCaptionTrackMenuItemText): Deleted.
(WebCore::sdhTrackKindDisplayName): Deleted.
(WebCore::sdhTrackMenuItemText): Deleted.
(WebCore::easyReaderKindDisplayName): Deleted.
(WebCore::easyReaderTrackMenuItemText): Deleted.
(WebCore::forcedTrackKindDisplayName): Deleted.
(WebCore::forcedTrackMenuItemText): Deleted.
(WebCore::audioDescriptionTrackSuffixText): Deleted.

• en.lproj/Localizable.strings:

Clean up localized string functions related to media controls.

LayoutTests:

• media/modern-media-controls/tracks-support/sorted-by-user-preferred-languages-expected.txt:

9:26 AM Changeset in webkit [278201] by commit-queue@webkit.org

• 8 edits in trunk/Source/WebCore

Unreviewed, reverting r278196.
​https://bugs.webkit.org/show_bug.cgi?id=226379

Causes asserts in layoutformattingcontext

Reverted changeset:

"[LFC] Transition containing block for loops to range loops"
​https://bugs.webkit.org/show_bug.cgi?id=226315
​https://trac.webkit.org/changeset/278196

9:24 AM Changeset in webkit [278200] by commit-queue@webkit.org

• 9 edits
  2 deletes in trunk

Unreviewed, reverting r278199.
​https://bugs.webkit.org/show_bug.cgi?id=226378

Causes asserts in layoutformattingcontext

Reverted changeset:

"[LFC][TFC] Do not include vertical spacing when resolving
percent height for table content"
​https://bugs.webkit.org/show_bug.cgi?id=226365
​https://trac.webkit.org/changeset/278199

8:59 AM Changeset in webkit [278199] by Alan Bujtas

• 9 edits
  2 adds in trunk

[LFC][TFC] Do not include vertical spacing when resolving percent height for table content
​https://bugs.webkit.org/show_bug.cgi?id=226365

Source/WebCore:

Reviewed by Antti Koivisto.

<table style="height: 100px; border-spacing: 10px;"><tr><td style="height: 100%"></td></tr></table>
The <td>'s 100% height is resolved against 100px - (2 * 10px) -> 80px;

Test: fast/layoutformattingcontext/table-with-percent-columns-and-spacing.html

• layout/formattingContexts/table/TableFormattingContext.cpp:

(WebCore::Layout::TableFormattingContext::TableFormattingContext):

• layout/formattingContexts/table/TableFormattingContext.h:
• layout/formattingContexts/table/TableFormattingGeometry.cpp:

(WebCore::Layout::TableFormattingGeometry::TableFormattingGeometry):
(WebCore::Layout::TableFormattingGeometry::computedCellBorder const):

• layout/formattingContexts/table/TableFormattingGeometry.h:

(WebCore::Layout::TableFormattingGeometry::formattingContext const):

• layout/formattingContexts/table/TableFormattingQuirks.cpp:

(WebCore::Layout::TableFormattingQuirks::heightValueOfNearestContainingBlockWithFixedHeight const):

• layout/formattingContexts/table/TableFormattingQuirks.h:

(WebCore::Layout::TableFormattingQuirks::formattingContext const):

LayoutTests:

Reviewed by Antti Koivisto.

• TestExpectations: WebKit table layout don't resolve the height percent properly when border spacing is present.
• fast/layoutformattingcontext/table-with-percent-columns-and-spacing-expected.html: Added.
• fast/layoutformattingcontext/table-with-percent-columns-and-spacing.html: Added.

8:56 AM Changeset in webkit [278198] by commit-queue@webkit.org

• 2 edits in trunk/Source/WebKit

[WPE] Correctly compute wheel event phase for 2D axis events
​https://bugs.webkit.org/show_bug.cgi?id=226370

Patch by Zan Dobersek <​zdobersek@igalia.com> on 2021-05-28
Reviewed by Adrian Perez de Castro.

2D-capable wpe_input_axis_event objects don't have usable axis and delta
values set on the base struct, but keep all that information on the more
detailed wpe_input_axis_2d_event struct.

For such events, the correct phase then has to be special-cased,
otherwise the default determination marks both axes as inactive and only
PhaseEnded events are dispatched into the engine.

• UIProcess/API/wpe/WPEView.cpp:

(WKWPE::m_backend):

8:50 AM Changeset in webkit [278197] by eric.carlson@apple.com

• 5 edits
  2 adds in trunk

[GPUP] [Video/Audio/Text]TrackPrivateRemote.h should use startTimeVariance
​https://bugs.webkit.org/show_bug.cgi?id=226355
rdar://77326202

Reviewed by Jer Noble.

Source/WebKit:

Test: http/tests/media/hls/hls-webvtt-seek-backwards.html

• WebProcess/GPU/media/AudioTrackPrivateRemote.h: Use configuration.startTimeVariance.
• WebProcess/GPU/media/TextTrackPrivateRemote.h: Ditto.
• WebProcess/GPU/media/VideoTrackPrivateRemote.h: Ditto.

LayoutTests:

• http/tests/media/hls/hls-webvtt-seek-backwards-expected.txt: Added.
• http/tests/media/hls/hls-webvtt-seek-backwards.html: Added.

7:20 AM Changeset in webkit [278196] by Alan Bujtas

• 8 edits in trunk/Source/WebCore

[LFC] Transition containing block for loops to range loops
​https://bugs.webkit.org/show_bug.cgi?id=226315

Reviewed by Antti Koivisto.

Add support for stayWithin.

• layout/floats/FloatingContext.cpp:

(WebCore::Layout::FloatingContext::mapTopLeftToFloatingStateRoot const):

• layout/formattingContexts/FormattingGeometry.cpp:

(WebCore::Layout::FormattingGeometry::computedHeightValue const):

• layout/formattingContexts/block/BlockFormattingContext.cpp:

(WebCore::Layout::BlockFormattingContext::usedAvailableWidthForFloatAvoider):

• layout/layouttree/LayoutBox.cpp:

(WebCore::Layout::Box::isDescendantOf const):

• layout/layouttree/LayoutBox.h:
• layout/layouttree/LayoutContainingBlockChainIterator.h:

(WebCore::Layout::LayoutContainingBlockChainIteratorAdapter::begin):
(WebCore::Layout::LayoutContainingBlockChainIteratorAdapter::end):
(WebCore::Layout::LayoutContainingBlockChainIterator::LayoutContainingBlockChainIterator):
(WebCore::Layout::LayoutContainingBlockChainIteratorAdapter::LayoutContainingBlockChainIteratorAdapter):
(WebCore::Layout::containingBlockChain):

6:23 AM Changeset in webkit [278195] by commit-queue@webkit.org

• 3 edits in trunk/Source/WebKit

[SOUP] Implement NetworkDataTask::setPriority
​https://bugs.webkit.org/show_bug.cgi?id=226371

Patch by Carlos Garcia Campos <​cgarcia@igalia.com> on 2021-05-28
Reviewed by Adrian Perez de Castro.

To update the soup message priority. In HTTP/2 soup can send a priority frame to prioritize the associated
stream.

• NetworkProcess/soup/NetworkDataTaskSoup.cpp:

(WebKit::NetworkDataTaskSoup::setPriority):

• NetworkProcess/soup/NetworkDataTaskSoup.h:

5:59 AM WebKitGTK/2.32.x edited by Adrian Perez de Castro

(diff)

4:35 AM Changeset in webkit [278194] by commit-queue@webkit.org

• 3 edits
  2 adds in trunk

Webkit doesn't calculate padding-top/-bottom in % correctly when parent has padding
​https://bugs.webkit.org/show_bug.cgi?id=221202

Patch by Rob Buis <​rbuis@igalia.com> on 2021-05-28
Reviewed by Manuel Rego Casasnovas.

LayoutTests/imported/w3c:

Add test for this.

• web-platform-tests/css/css-position/position-absolute-padding-percentage-expected.txt: Added.
• web-platform-tests/css/css-position/position-absolute-padding-percentage.html: Added.

Source/WebCore:

Absolute positioned elements should resolve %-ge padding against containing block padding-edge:
​https://drafts.csswg.org/css-position-3/#absolute-positioning-containing-block

Test: imported/w3c/web-platform-tests/css/css-position/position-absolute-padding-percentage.html

• rendering/RenderBox.cpp:

(WebCore::RenderBox::containingBlockLogicalWidthForContent const):

1:59 AM Changeset in webkit [278193] by Martin Robinson

• 8 edits
  2 adds in trunk

More readily layout when scroll-snap properties change
​https://bugs.webkit.org/show_bug.cgi?id=225950

Reviewed by Simon Fraser.

LayoutTests/imported/w3c:

• web-platform-tests/css/css-scroll-snap/scroll-snap-type-change-expected.txt: Updated expectation

for newly passing test.

Source/WebCore:

Do a layout when scroll-snap properties change instead of trying
to update scroll positions without a layout.

Test: css3/scroll-snap/scroll-snap-style-change-crash.html

• rendering/RenderLayerModelObject.cpp:

(WebCore::RenderLayerModelObject::styleDidChange): Remove code dealing with scrollports
as this is now handled by the RenderStyle diff. Now trigger a layout for scrollports
when the scrolled children change their snap properties.

• rendering/RenderObject.cpp:

(WebCore::RenderObject::enclosingScrollableContainerForSnapping const): Made this return a non-const.

• rendering/RenderObject.h:
• rendering/style/RenderStyle.cpp:

(WebCore::RenderStyle::changeRequiresLayout const): Added code to handle scrollport specific
properties.

LayoutTests:

• css3/scroll-snap/scroll-snap-style-change-crash-expected.txt: Added.
• css3/scroll-snap/scroll-snap-style-change-crash.html: Added.

12:15 AM Changeset in webkit [278192] by commit-queue@webkit.org

• 2 edits in trunk/Source/ThirdParty/ANGLE

ANGLE Metal translator should always collect variables
​https://bugs.webkit.org/show_bug.cgi?id=226261

Patch by Kimmo Kinnunen <​kkinnunen@apple.com> on 2021-05-28
Reviewed by Dean Jackson.

Always collect the variables when translating GLSL to Metal.
This way tests that invoke translation without SH_VARIABLES still
work.

• src/compiler/translator/TranslatorMetalDirect.cpp:

(sh::TranslatorMetalDirect::translateImpl):

• src/compiler/translator/TranslatorMetalDirect.h:

May 27, 2021:

10:26 PM Changeset in webkit [278191] by Wenson Hsieh

• 7 edits in trunk/Source/WebKit

[GPU Process] Add debug logging around some shared display list codepaths
​https://bugs.webkit.org/show_bug.cgi?id=226356

Reviewed by Simon Fraser.

Add logging around the shared display list processing logic between the web and GPU processes. This is the basic
set of log messages I've used in the past to triage crashes and hangs while optimizing MotionMark performance,
and should be sufficient for some basic debugging around display list code around RemoteRenderingBackend(Proxy).

• GPUProcess/graphics/RemoteImageBuffer.h:
• GPUProcess/graphics/RemoteRenderingBackend.cpp:

(WebKit::RemoteRenderingBackend::nextDestinationImageBufferAfterApplyingDisplayLists):
(WebKit::RemoteRenderingBackend::wakeUpAndApplyDisplayList):

Log when wakeup messages are sent by the web process and received by the GPU process.

• GPUProcess/graphics/RemoteRenderingBackend.h:
• Platform/Logging.h:
• WebProcess/GPU/graphics/RemoteImageBufferProxy.h:

(WebKit::RemoteImageBufferProxy::waitForDidFlushWithTimeout):

Log when synchronous flushes are performed, as well as whether they time out.

• WebProcess/GPU/graphics/RemoteRenderingBackendProxy.cpp:

(WebKit::RemoteRenderingBackendProxy::sendWakeupMessage):
(WebKit::RemoteRenderingBackendProxy::createItemBuffer):

Log whenever new display list shared memory is created, and also when display lists are reused (additionally
logging the remaining capacity on reused display lists).

10:10 PM Changeset in webkit [278190] by Wenson Hsieh

• 23 edits
  3 adds in trunk/Source

[macOS] Show context menu when clicking on data detection results in image overlays
​https://bugs.webkit.org/show_bug.cgi?id=226274
rdar://75504956

Reviewed by Tim Horton and Aditya Keerthi.

Source/WebCore:

Add a client hook to allow ImageOverlayController to call out into the WebKit client layer when a data detector
button (with a corresponding DDScannerResult) is clicked. See WebKit ChangeLog for more details.

• WebCore.xcodeproj/project.pbxproj:
• page/ChromeClient.h:

(WebCore::ChromeClient::handleClickForDataDetectionResult):

• page/cocoa/DataDetectorElementInfo.h: Added.

Introduce a struct that wraps a data detection result and the accompanying element bounding rect, mainly so that
we have a C++ struct we can plumb through the client layers (rather than directly passing Objective-C objects
around).

• page/mac/ImageOverlayControllerMac.mm:

(WebCore::ImageOverlayController::handleDataDetectorAction):

Source/WebCore/PAL:

Add declarations for some SPI on RVItem and RVPresenter. See WebKit for more information about where these are
used.

• pal/spi/cocoa/RevealSPI.h:

Source/WebKit:

Implement a WebKit2 client hook to handle a click over a data detection result by using the Reveal framework to
collect and present a context menu. See below for more details.

• Shared/Cocoa/WebCoreArgumentCodersCocoa.mm:

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

• Shared/WebCoreArgumentCoders.h:

Add encoding and decoding support for DataDetectorElementInfo.

• SourcesCocoa.txt:
• UIProcess/Cocoa/WebPageProxyCocoa.mm:

(WebKit::WebPageProxy::handleClickForDataDetectionResult):

• UIProcess/Cocoa/WebViewImpl.h:

(WebKit::WebViewImpl::view const):

Add a helper method to grab the view (either a WKView or WKWebView) from a WebViewImpl.

• UIProcess/Cocoa/WebViewImpl.mm:

(WebKit::WebViewImpl::handleClickForDataDetectionResult):

Create and present a new WKRevealItemPresenter using the DDScannerResult.

(WebKit::WebViewImpl::didFinishPresentation):

• UIProcess/PageClient.h:

(WebKit::PageClient::handleClickForDataDetectionResult):

Add more plumbing through the UI process to handle clicking on a data detection result.

• UIProcess/WebPageProxy.h:
• UIProcess/WebPageProxy.messages.in:
• UIProcess/mac/PageClientImplMac.h:
• UIProcess/mac/PageClientImplMac.mm:

(WebKit::PageClientImpl::handleClickForDataDetectionResult):

• UIProcess/mac/WKRevealItemPresenter.h: Added.
• UIProcess/mac/WKRevealItemPresenter.mm: Added.

Add a helper class that takes an RVItem and is capable of presenting a context menu for the item, on behalf of
WebViewImpl. For the time being, this is only used on macOS (and thus, directly holds a WeakPtr to a
WebViewImpl). However, we may want to use this in the future for Mac Catalyst as well, in which case we should
make this work via a generic WKRevealItemPresenterDelegate protocol instead.

(-[WKRevealItemPresenter initWithWebViewImpl:item:frame:menuLocation:]):
(-[WKRevealItemPresenter showContextMenu]):
(-[WKRevealItemPresenter _callDidFinishPresentationIfNeeded]):

Call out to the WebViewImpl to reset its current reveal item presenter after we've finished presentation. This
happens either after we've dismissed the context menu (if no item is highlighted), or after we've finished
highlighting the item.

(-[WKRevealItemPresenter revealContext:rectsForItem:]):
(-[WKRevealItemPresenter revealContext:shouldUseDefaultHighlightForItem:]):
(-[WKRevealItemPresenter revealContext:startHighlightingItem:]):
(-[WKRevealItemPresenter revealContext:stopHighlightingItem:]):

• WebKit.xcodeproj/project.pbxproj:
• WebProcess/WebCoreSupport/WebChromeClient.cpp:

(WebKit::WebChromeClient::handleClickForDataDetectionResult):

• WebProcess/WebCoreSupport/WebChromeClient.h:
• WebProcess/WebPage/Cocoa/WebPageCocoa.mm:

(WebKit::WebPage::handleClickForDataDetectionResult):

• WebProcess/WebPage/WebPage.h:

10:02 PM Changeset in webkit [278189] by Devin Rousso

• 4 edits in trunk

Sampled Page Top Color: allow snapshotting of elements with CSS animations/transitions if they're paused
​https://bugs.webkit.org/show_bug.cgi?id=226313
<rdar://problem/78534076>

Reviewed by Tim Horton.

Source/WebCore:

The primary concern of sampling elements with CSS animations/transitions is that it
introduces some indeterminism in when the sampling happens vs how far progressed the CSS
animation/transition is. As an example, sampling from a page with an inline <style> that
applies a CSS animation to fade in the CSS background-color from white to black is
more likely to get a color closer to black than if that same CSS was in an external
uncached resource. This wouldn't make for a great experience, so r277044 made it so that any
CSS animations/transitions caused the sampling logic to bail, regardless of the state of the
CSS animation/transition. This is only really an issue for CSS animations/transitions that
are actively running, however, not ones that have yet to run or have already finished. It's
still possible that two loads of the same page could result in different colors (or bails)
depending on how quickly the CSS animation/transition runs or (if the CSS is in an external
resource) how long the containing resource takes to load, but with this patch it's now a
binary state (sample or bail) instead of an indeterminate range of possible sampled colors.

Tests: SampledPageTopColor.HitTestBeforeCSSTransition

SampledPageTopColor.HitTestDuringCSSTransition
SampledPageTopColor.HitTestAfterCSSTransition
SampledPageTopColor.HitTestBeforeCSSAnimation
SampledPageTopColor.HitTestDuringCSSAnimation
SampledPageTopColor.HitTestAfterCSSAnimation

• page/PageColorSampler.cpp:

(WebCore::isValidSampleLocation):
Use Styleable instead of RenderStyle to get more information about active CSS animations/transitions.

Tools:

• TestWebKitAPI/Tests/WebKitCocoa/SampledPageTopColor.mm:

(waitForSampledPageTopColorToChange): Added.
(waitForSampledPageTopColorToChangeForHTML):
(TEST.SampledPageTopColor.HitTestBeforeCSSTransition): Added.
(TEST.SampledPageTopColor.HitTestDuringCSSTransition): Added.
(TEST.SampledPageTopColor.HitTestAfterCSSTransition): Added.
(TEST.SampledPageTopColor.HitTestBeforeCSSAnimation): Added.
(TEST.SampledPageTopColor.HitTestDuringCSSAnimation): Added.
(TEST.SampledPageTopColor.HitTestAfterCSSAnimation): Added.
(TEST.SampledPageTopColor.HitTestCSSAnimation): Deleted.

9:16 PM Changeset in webkit [278188] by Said Abou-Hallawa

• 6 edits
  2 adds in trunk

Values of keySplines control points must all be in the range 0 to 1
​https://bugs.webkit.org/show_bug.cgi?id=226336

Reviewed by Sam Weinig.

LayoutTests/imported/w3c:

• web-platform-tests/svg/animations/keysplines-x-limits-expected.txt:
• web-platform-tests/svg/animations/keysplines-y-limits-expected.txt: Added.
• web-platform-tests/svg/animations/keysplines-y-limits.html: Added.

Source/WebCore:

W3C specs:

​https://svgwg.org/specs/animations/#KeySplinesAttribute

MDN documenation:

​https://developer.mozilla.org/en-US/docs/Web/SVG/Attribute/keySplines

Test: imported/w3c/web-platform-tests/svg/animations/keysplines-y-limits.html

• svg/SVGAnimationElement.cpp:

(WebCore::parseKeySplines):

Source/WTF:

• wtf/MathExtras.h:

(isInRange):
There are a least couple of implementations of this function in WebCore.
The next step is to remove the redundant code in WebCore and use this one.

7:26 PM Changeset in webkit [278187] by Jonathan Bedard

• 4 edits in trunk/Tools

[webkitcorey] Gracefully handle CNTRL-C in TaskPool
​https://bugs.webkit.org/show_bug.cgi?id=226238
<rdar://problem/78472148>

Reviewed by Dewei Zhu.

• Scripts/libraries/webkitcorepy/setup.py: Bump version.
• Scripts/libraries/webkitcorepy/webkitcorepy/init.py: Ditto.
• Scripts/libraries/webkitcorepy/webkitcorepy/task_pool.py:

(_BiDirectionalQueue.close): Tear-down queue objects while suppressing any logging.
(_Process.handler): Both SIGTERM and SIGINT signals should stop worker processes.
(_Process.main): Add SIGINT handler, explicitly close queue.
(TaskPool.init): Defer worker and queue construction to context manager so that we
do not have an instantiated but invalid queue if pipes are broken by children processes.
(TaskPool.enter): Construct worker processes.
(TaskPool.do): Only catch Queue.Empty exception.
(TaskPool.exit): Explicitly close queue, reset workers and queue.

7:08 PM Changeset in webkit [278186] by rmorisset@apple.com

• 8 edits
  1 add in trunk/Source

AirAllocateStackByGraphColoring should use the optimized interference graphs from AirAllocateRegistersByGraphColoring
​https://bugs.webkit.org/show_bug.cgi?id=226258

Reviewed by Phil Pizlo.

Source/JavaScriptCore:

The main change in this patch is that AirAllocateStackByGraphColoring is now using the optimized datastructures in wtf/InterferenceGraph.h.
This required templating most of it over the interference graph used (Small/Large/Huge), but I tried keeping some common parts out of the templated class to minimize the impact on compile times and binary size.

A consequence of that change is that coalescableMoves and remappedStackSlots now store indices instead of direct pointers to StackSlots, resulting in a memory reduction of about 3x as well.

Another consequence is that I had to slightly alter the way that coalescing works: instead of carefully removing the interference edges of the killed slot, we simply use mayClear() which is not guaranteed to remove anything.
I believe that this is sound, because every subsequent access to m_interference checks whether a slot has been coalesced first, so dropping these edges is purely a memory saving, but has no logical effect.

The new code was tested in a few ways:

• running on JetStream2 with asan
• running on JetStream2 with TEST_OPTIMIZED_INTERFERENCE_GRAPH
• running on JetStream2 and logging the frame sizes at the end of this phase, and comparing to the results of doing the same on ToT (same average frame size)

The two functions where this code had the largest memory footprint in JetStream2 were both in tsf-wasm.
One has 751 stack slots, and had an interference graph of 2.1MB and a coalescableMoves vector of 440kB
The other has 673 stack slots, and had an interference graph of 1.9MB and a coalescableMoves vector of 421kB.
With this patch, they respectively use 79kB+146kB and 67kB+140kB
The effect on the rest of JetStream2 is less likely to matter as few functions used more than a few dozens of kB in this phase, but in percentages are just as huge.

More importantly (and the reason I wrote this patch in the first place), I checked mruby-wasm.aotoki.dev which with a few other pages forced us to lower Options::maximumTmpsForGraphColoring because of jetsams.
It has two massive functions that reach this phase if I increase Options::maximumTmpsForGraphColoring:

• about 6k stack slots -> 215MB + 6MB (interference graph + coalescableMoves)
• about 9k stack slots -> 395MB + 4MB

After this patch, they respectively use 4.5MB+2MB and 9MB+1.5MB, or roughly a 40x improvement.
Combined with the recent improvements to the register allocator, I hope to be able to increase Options::maximumTmpsForGraphColoring soon (in a different patch for easier bisection if either cause a perf regression).
This would be helpful, since its lowering cratered our performance on some other wasm application by 8x.

In terms of compile times, this patch lowered the time spent in AllocateStackByGraphColoring over the course of a run of JetStream2 from roughly 350ms to roughly 270ms.
This is almost certainly negligible, but at least it guarantees that it did not regress.

• b3/air/AirAllocateRegistersByGraphColoring.cpp:
• b3/air/AirAllocateStackByGraphColoring.cpp:

(JSC::B3::Air::allocateStackByGraphColoring):

Source/WTF:

I moved the interference graphs datastructures from AirAllocateRegistersByGraphColoring to their own wtf/InterferenceGraph.h file.
There are three of them:

• SmallInterferenceGraph, best for n < 400
• LargeInterferenceGraph, for n < 216
• HugeInterferenceGraph, for n up to 232

I also added "Iterable" versions of them, that have an operator[] method whose result you can iterate on to get all the indices which interfere with a given index.
SmallIterableInterferenceGraph is the same as the non-iterable version, but the Large and Huge versions are a bit slower than their counterparts and use 2x memory.

All of these were tested by running JetStream2 with the TEST_OPTIMIZED_INTERFERENCE_GRAPH set to 1.
This flag makes the optimized datastructures run in parallel with a reference implementation, and their results are checked for equality on every method call.
There is one small difference allowed: iteration is not guaranteed to go through elements in the same order.

I also added a clear() method to LikelyDenseUnsignedIntegerSet, and added the NotNull flag to its various uses of placement new.

• WTF.xcodeproj/project.pbxproj:
• wtf/CMakeLists.txt:
• wtf/HashSet.h:

(WTF::W>::memoryUse const):

• wtf/InterferenceGraph.h: Added.

(WTF::InterferenceBitVector::contains):
(WTF::InterferenceBitVector::addAndReturnIsNewEntry):
(WTF::InterferenceBitVector::add):
(WTF::InterferenceBitVector::clear):
(WTF::InterferenceBitVector::mayClear):
(WTF::InterferenceBitVector::setMaxIndex):
(WTF::InterferenceBitVector::forEach):
(WTF::InterferenceBitVector::size const):
(WTF::InterferenceBitVector::memoryUse const):
(WTF::InterferenceBitVector::dumpMemoryUseInKB const):
(WTF::InterferenceBitVector::Iterable::iterator::operator++):
(WTF::InterferenceBitVector::Iterable::iterator::operator* const):
(WTF::InterferenceBitVector::Iterable::iterator::operator== const):
(WTF::InterferenceBitVector::Iterable::iterator::operator!= const):
(WTF::InterferenceBitVector::Iterable::begin const):
(WTF::InterferenceBitVector::Iterable::end const):
(WTF::InterferenceBitVector::operator[] const):
(WTF::InterferenceBitVector::index const):
(WTF::InterferenceVector::contains):
(WTF::InterferenceVector::addAndReturnIsNewEntry):
(WTF::InterferenceVector::add):
(WTF::InterferenceVector::clear):
(WTF::InterferenceVector::mayClear):
(WTF::InterferenceVector::setMaxIndex):
(WTF::InterferenceVector::forEach):
(WTF::InterferenceVector::size const):
(WTF::InterferenceVector::memoryUse const):
(WTF::InterferenceVector::dumpMemoryUseInKB const):
(WTF::InterferenceVector::Iterable::begin const):
(WTF::InterferenceVector::Iterable::end const):
(WTF::InterferenceVector::operator[] const):
(WTF::UndirectedEdgesDuplicatingAdapter::contains):
(WTF::UndirectedEdgesDuplicatingAdapter::addAndReturnIsNewEntry):
(WTF::UndirectedEdgesDuplicatingAdapter::add):
(WTF::UndirectedEdgesDuplicatingAdapter::clear):
(WTF::UndirectedEdgesDuplicatingAdapter::mayClear):
(WTF::UndirectedEdgesDuplicatingAdapter::setMaxIndex):
(WTF::UndirectedEdgesDuplicatingAdapter::forEach):
(WTF::UndirectedEdgesDuplicatingAdapter::size const):
(WTF::UndirectedEdgesDuplicatingAdapter::memoryUse const):
(WTF::UndirectedEdgesDuplicatingAdapter::dumpMemoryUseInKB const):
(WTF::UndirectedEdgesDuplicatingAdapter::operator[] const):
(WTF::UndirectedEdgesDedupAdapter::contains):
(WTF::UndirectedEdgesDedupAdapter::addAndReturnIsNewEntry):
(WTF::UndirectedEdgesDedupAdapter::add):
(WTF::UndirectedEdgesDedupAdapter::clear):
(WTF::UndirectedEdgesDedupAdapter::mayClear):
(WTF::UndirectedEdgesDedupAdapter::setMaxIndex):
(WTF::UndirectedEdgesDedupAdapter::forEach):
(WTF::UndirectedEdgesDedupAdapter::size const):
(WTF::UndirectedEdgesDedupAdapter::memoryUse const):
(WTF::UndirectedEdgesDedupAdapter::dumpMemoryUseInKB const):
(WTF::InterferenceHashSet::contains):
(WTF::InterferenceHashSet::addAndReturnIsNewEntry):
(WTF::InterferenceHashSet::add):
(WTF::InterferenceHashSet::clear):
(WTF::InterferenceHashSet::setMaxIndex):
(WTF::InterferenceHashSet::forEach):
(WTF::InterferenceHashSet::size const):
(WTF::InterferenceHashSet::memoryUse const):
(WTF::InterferenceHashSet::dumpMemoryUseInKB const):
(WTF::InstrumentedInterferenceGraph::contains):
(WTF::InstrumentedInterferenceGraph::addAndReturnIsNewEntry):
(WTF::InstrumentedInterferenceGraph::add):
(WTF::InstrumentedInterferenceGraph::clear):
(WTF::InstrumentedInterferenceGraph::mayClear):
(WTF::InstrumentedInterferenceGraph::setMaxIndex):
(WTF::InstrumentedInterferenceGraph::forEach):
(WTF::InstrumentedInterferenceGraph::size const):
(WTF::InstrumentedInterferenceGraph::memoryUse const):
(WTF::InstrumentedInterferenceGraph::dumpMemoryUseInKB const):
(WTF::InstrumentedIterableInterferenceGraph::Iterable::Iterable):
(WTF::InstrumentedIterableInterferenceGraph::Iterable::begin const):
(WTF::InstrumentedIterableInterferenceGraph::Iterable::end const):
(WTF::InstrumentedIterableInterferenceGraph::operator[] const):

• wtf/LikelyDenseUnsignedIntegerSet.h:

(WTF::LikelyDenseUnsignedIntegerSet::LikelyDenseUnsignedIntegerSet):
(WTF::LikelyDenseUnsignedIntegerSet::clear):
(WTF::LikelyDenseUnsignedIntegerSet::add):
(WTF::LikelyDenseUnsignedIntegerSet::estimateHashSetSize):
(WTF::LikelyDenseUnsignedIntegerSet::transitionToHashSet):
(WTF::LikelyDenseUnsignedIntegerSet::transitionToBitVector):

6:26 PM Changeset in webkit [278185] by Darin Adler

• 1470 edits in trunk

Next step toward using std::optional directly instead of through WTF::Optional typedef
​https://bugs.webkit.org/show_bug.cgi?id=226280

Reviewed by Chris Dumez.

Source/JavaScriptCore:

• <many files>: Accept the renaming done by do-webcore-rename.

• yarr/YarrSyntaxChecker.cpp: Since the style checker complained about this file,

tweaked style to make it happy after the renaming done by do-webcore-rename, and
also hand-updated Optional to std::optional as long as we were touching it.

Source/WebCore:

• <many files>: Accept the renaming done by do-webcore-rename.

• Modules/webauthn/fido/DeviceRequestConverter.h: Since style checker complained

about the names of some arguments, fixed them, and also hand-updated Optional to
std::optional as long as we were touching it.

• loader/EmptyClients.cpp: Since style checker complained about the mix of

WEBCORE_EXPORT and inlined functions, moved them out of line, and
also hand-updated Optional to std::optional as long as we were touching it.
Also removed is<EmptyFrameLoaderClient>().

• loader/EmptyFrameLoaderClient.h: Ditto.

Source/WebCore/PAL:

• <many files>: Accept the renaming done by do-webcore-rename.

Source/WebDriver:

• <many files>: Accept the renaming done by do-webcore-rename.

Source/WebKit:

• <many files>: Accept the renaming done by do-webcore-rename.

Source/WebKitLegacy:

• Storage/StorageTracker.cpp:

(WebKit::StorageTracker::diskUsageForOrigin): Accept the renaming done by do-webcore-rename.

Source/WebKitLegacy/mac:

• <many files>: Accept the renaming done by do-webcore-rename.

Source/WebKitLegacy/win:

• <many files>: Accept the renaming done by do-webcore-rename.

Source/WTF:

• <many files>: Accept the renaming done by do-webcore-rename.

• wtf/Optional.h: Remove WTF::nullopt_t and WTF::makeOptional.

• wtf/URLHelpers.cpp:

(WTF::URLHelpers::mapHostName): Convert from nullopt to std::nullopt.

Tools:

• Scripts/do-webcore-rename: Use script to rename valueOr, WTF::nullopt, WTF::nullopt_t,

WTF::Optional, WTF::makeOptional, and makeOptional. Other renamings can't necessarily
be done by the script and so will be done in later passes.

• <many files>: Accept the renaming done by do-webcore-rename.

6:07 PM Changeset in webkit [278184] by Jonathan Bedard

• 3 edits in trunk/Tools

[webkitpy] Use existing signal handler when printing stacktrace
​https://bugs.webkit.org/show_bug.cgi?id=226236
<rdar://problem/78471882>

Reviewed by Dewei Zhu.

• Scripts/webkitpy/common/interrupt_debugging.py:

(log_stack_trace_on_signal): Use the same implementation for SIGTERM and SIGINT signals, call existing
handler after logging stack trace.
(log_stack_trace_on_term): Deleted.
(log_stack_trace_on_ctrl_c): Deleted.

• Scripts/webkitpy/layout_tests/run_webkit_tests.py:

(main): Use the new log_stack_trace_on_signal function.

5:40 PM Changeset in webkit [278183] by Devin Rousso

• 7 edits in trunk/Source/WebKit

[iOS] contextmenu hints don't follow scrolling inside <iframe>
​https://bugs.webkit.org/show_bug.cgi?id=226055

Reviewed by Tim Horton.

• UIProcess/ios/WKContentViewInteraction.h:
• UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView _didScroll]):
(-[WKContentView _removeContainerForContextMenuHintPreviews]):
(-[WKContentView _updateFrameOfContainerForContextMenuHintPreviewsIfNeeded]): Added.
(-[WKContentView _updateTargetedPreviewScrollViewUsingContainerScrollingNodeID:]): Added.
(-[WKContentView _createTargetedContextMenuHintPreviewForFocusedElement]): Added.
(-[WKContentView _createTargetedContextMenuHintPreviewIfPossible]): Added.
(-[WKContentView overridePositionTrackingViewForTargetedPreviewIfNecessary:containerScrollingNodeID:]): Deleted.
Save the UIScrollView containing the target of the contextmenu. Whenever any view is
scrolled, adjust the frame of the _contextMenuHintContainerView so that the contextmenu
hint moves by the amount of scrolling that's happened since the contextmenu hint was shown.

• UIProcess/RemoteLayerTree/ios/RemoteScrollingCoordinatorProxyIOS.mm:

(WebKit::RemoteScrollingCoordinatorProxy::scrollViewForScrollingNodeID const):

• UIProcess/RemoteLayerTree/ios/ScrollingTreeFrameScrollingNodeRemoteIOS.h:
• UIProcess/RemoteLayerTree/ios/ScrollingTreeFrameScrollingNodeRemoteIOS.mm:

(WebKit::ScrollingTreeFrameScrollingNodeRemoteIOS::scrollView const): Added.
ScrollingTreeFrameScrollingNodeRemoteIOS also has a corresponding UIScrollView.

• Platform/spi/ios/UIKitSPI.h:

Remove declaration for no-longer-used SPI.

4:29 PM Changeset in webkit [278182] by Amir Mark Jr.

• 2 edits in trunk/LayoutTests

[MacOS Wk1] imported/w3c/web-platform-tests/mathml/presentation-markup/mrow/inferred-mrow-stretchy.html is a flaky failure
​https://bugs.webkit.org/show_bug.cgi?id=226360

Unreviewed test gardening.

• platform/mac-wk1/TestExpectations:

4:07 PM Changeset in webkit [278181] by Brent Fulgham

• 2 edits in trunk/Source/WebKit

[iOS] Silence reports about 'com.apple.runningboard'
​https://bugs.webkit.org/show_bug.cgi?id=226357
<rdar://problem/76144219>

Reviewed by Per Arne Vollan.

We blocked access to 'com.apple.runningboard' from the WebContent process in Bug 209933 (about a year ago). Since
then, the only telemetry we have received has been for a benign Accessibility call that fails closed in way that
does not impact the features of the Accessibility Framework used inside the WebContent process.

We should silence this report to avoid spending time symbolicating and generating traces.

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

4:03 PM Changeset in webkit [278180] by Brent Fulgham

• 2 edits in trunk/Source/WebCore

[Hardening] Use a RELEASE_ASSERT for SharedBufferDataView constructor
​https://bugs.webkit.org/show_bug.cgi?id=226346
<rdar://problem/78115298>

Reviewed by Ryosuke Niwa.

We have a DEBUG assert in the constructor for SharedBufferDataView. Debug test show that we don't hit this during normal
operation, and it's not a hot code path. We should make it a RELEASE_ASSERT.

• platform/SharedBuffer.cpp:

(WebCore::SharedBufferDataView::SharedBufferDataView):

2:55 PM Changeset in webkit [278179] by sihui_liu@apple.com

• 9 edits in trunk/Source

Abandon pending tasks on background thread when WebIDBServer is closed
​https://bugs.webkit.org/show_bug.cgi?id=226295

Reviewed by Chris Dumez.

Source/WebCore:

• Modules/indexeddb/server/SQLiteIDBBackingStore.cpp: Disable threading check as we may

abort transactions on the main thread.
(WebCore::IDBServer::SQLiteIDBBackingStore::getOrEstablishDatabaseInfo):

Source/WebKit:

When WebIDBServer is closed, it means there is no open connection to client, so we don't need to handle pending
tasks. We can ditch them by killing the CrossThreadQueue. For ongoing task, it can be blocked on quota check, so
let's just stop database activities on the main thread to make sure database lock is released.

• NetworkProcess/IndexedDB/WebIDBServer.cpp:

(WebKit::WebIDBServer::close):

• NetworkProcess/IndexedDB/WebIDBServer.h:

Source/WTF:

Make the wait end with either a new message or queue being killed.

• wtf/CrossThreadQueue.h:

(WTF::CrossThreadQueue<DataType>::waitForMessage):

• wtf/CrossThreadTask.h:

(WTF::CrossThreadTask::operator bool const):

• wtf/CrossThreadTaskHandler.cpp:

(WTF::CrossThreadTaskHandler::taskRunLoop):

2:35 PM Changeset in webkit [278178] by don.olmstead@sony.com

• 6 edits in trunk

[CMake] Add check for timingsafe_bcmp
​https://bugs.webkit.org/show_bug.cgi?id=226347

Reviewed by Chris Dumez.

.:

Adds a symbol check for timingsafe_bcmp and sets HAVE_TIMINGSAFE_BCMP accordingly in the
cmakeconfig.h.

• Source/cmake/OptionsCommon.cmake:

Source/WTF:

Use HAVE(TIMINGSAFE_BCMP) in CryptographicUtilities code. Set its value in PlatformHave for
platforms not using CMake.

• wtf/CryptographicUtilities.cpp:
• wtf/CryptographicUtilities.h:
• wtf/PlatformHave.h:

2:34 PM Changeset in webkit [278177] by commit-queue@webkit.org

• 7 edits in trunk

Increase NumberToStringBuffer to account for negative number
​https://bugs.webkit.org/show_bug.cgi?id=226014

Patch by Mikhail R. Gadelha <​mikhail.ramalho@gmail.com> on 2021-05-27
Reviewed by Robin Morisset.

In r250389, NumberToStringBuffer length was increased from 96 to 123 because:

<21 digits> + decimal point + <100 digits> + null char = 123.

however,

a = -100000000000000000000
a.toFixed(100)

is a valid conversion and it is 124 chars long, because of the "-" sign.

So this patch increases the length of NumberToStringBuffer to 124 and adds new
test cases.

Source/WTF:

Reviewed by Ryosuke Niwa.

• wtf/dtoa.h:

LayoutTests:

• js/dom/number-tofixed-expected.txt:
• js/dom/script-tests/number-tofixed.js:
• js/kde/Number-expected.txt:
• js/kde/script-tests/Number.js:

1:48 PM Changeset in webkit [278176] by achristensen@apple.com

• 4 edits in trunk/Source

Revert r277344
​https://bugs.webkit.org/show_bug.cgi?id=225602
<rdar://78411776>

Source/WebKit:

It caused more sandbox issues. We are going to see if we can just remove the connection killing, but first we revert the cause of problems.

• UIProcess/Launcher/mac/ProcessLauncherMac.mm:

(WebKit::terminateWithReason):

Source/WTF:

• wtf/PlatformHave.h:

1:25 PM Changeset in webkit [278175] by Wenson Hsieh

• 11 edits
  1 copy
  1 add in trunk/Source

Refactor some data detector code in WebCore::Frame
​https://bugs.webkit.org/show_bug.cgi?id=226267

Reviewed by Andy Estes and Devin Rousso.

Source/WebCore:

In preparation for supporting data detection in image overlays, this patch refactors some existing data detector
logic inside (and adjacent to) Frame. See comments below for more details.

• WebCore.xcodeproj/project.pbxproj:
• editing/cocoa/DataDetection.h:
• editing/cocoa/DataDetection.mm:

(WebCore::DataDetection::canPresentDataDetectorsUIForElement):

Use DataDetectionResultsStorage instead of grabbing the array of scanner results directly from Frame.

(WebCore::DataDetection::createElementForImageOverlay):

Add a helper method to create a data detector element for image overlays; this grabs a scanner result from the
given ImageExtractionDataDetectorInfo, adds it to the element's document's frame, and then sets the newly
created element's "x-apple-data-detectors-result" attribute to the new identifier value. The main reason we do
this is to avoid having HTMLElement directly handle DDScannerResult (and instead delegate out to platform-
specific helpers in DataDetection).

• html/HTMLElement.cpp:

(WebCore::HTMLElement::updateWithImageExtractionResult):

Use the new helper function above.

• loader/FrameLoader.cpp:

(WebCore::FrameLoader::checkLoadCompleteForThisFrame):

Use DataDetectionResultsStorage instead of setting the array of scanner results directly on Frame.

• page/Frame.cpp:

(WebCore::Frame::dataDetectionResults):

• page/Frame.h:

We currently store a list of document-wide data detection results (Objective-C DDScannerResult objects)
directly on Frame. For image overlays, I need to store a map of identifiers to scanner results in a similar
fashion. Rather than add more Objective-C code on Frame, it's better to instead move this logic out into a
separate C++ class, DataDetectionResultsStorage, that wraps both the document-level array of DDScannerResults,
as well as the new map of ImageOverlayDataDetectionResultIdentifier => DDScannerResult specifically for data
detector results inside image overlays.

• page/cocoa/DataDetectionResultsStorage.h: Copied from Source/WebCore/editing/cocoa/DataDetection.h.

(WebCore::DataDetectionResultsStorage::setDocumentLevelResults):
(WebCore::DataDetectionResultsStorage::documentLevelResults const):

Move the existing m_dataDetectionResults NSArray on Frame out into the storage helper class, as
m_documentLevelResults. As its name suggests, this array stores data detection results gathered by running
data detection over the entire document.

(WebCore::DataDetectionResultsStorage::imageOverlayDataDetectionResult):
(WebCore::DataDetectionResultsStorage::addImageOverlayDataDetectionResult):

Add a new map of ImageOverlayDataDetectionResultIdentifier to DDScannerResult for storing scanner results
specific to image overlay content. Using a separate data structure here prevents image overlay data detector
results from getting blown away when document-level data detection finishes and sets document-wide results on
the frame.

• page/cocoa/ImageOverlayDataDetectionResultIdentifier.h: Added.

Add an identifier type that we can use to look up data detection results in image overlay elements in Frame's
data detection result storage.

Source/WebKit:

See WebCore/ChangeLog for more details.

• WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::removeDataDetectedLinks):
(WebKit::WebPage::detectDataInAllFrames):

Use DataDetectionResultsStorage instead of setting the array of scanner results directly on Frame.

• WebProcess/WebPage/ios/WebPageIOS.mm:

(WebKit::dataDetectorLinkPositionInformation):

1:06 PM Changeset in webkit [278174] by Fujii Hironori

• 4 edits
  1 move
  2 deletes in trunk

[run-webkit-tests][Win] Strip PHP support from webserver
​https://bugs.webkit.org/show_bug.cgi?id=226287

Reviewed by Jonathan Bedard.

Tools:

We don't have any PHP tests now.

• Scripts/webkitpy/port/base.py:

(Port._apache_version):
(Port._apache_config_file_name_for_platform):
(Port._win_php_version): Deleted.

• Scripts/webkitpy/port/port_testcase.py:

LayoutTests:

• http/conf/libphp4.dll: Removed.
• http/conf/win-httpd-2.4-php5.conf: Removed.
• http/conf/win-httpd-2.4.conf: Renamed from LayoutTests/http/conf/win-httpd-2.4-php7.conf.

12:51 PM Changeset in webkit [278173] by commit-queue@webkit.org

• 5 edits in trunk

[css-grid] Add discrete animation support for grid-template-columns|rows
​https://bugs.webkit.org/show_bug.cgi?id=226174

LayoutTests/imported/w3c:

Patch by Ziran Sun <Ziran Sun> on 2021-05-27
Reviewed by Antoine Quint.

• web-platform-tests/css/css-grid/animation/grid-template-columns-interpolation-expected.txt:
• web-platform-tests/css/css-grid/animation/grid-template-rows-interpolation-expected.txt:

Source/WebCore:

For CSSGridTemplateTemplate|Row, the animation wrapper setter needs to check a few style
settings such as gridColumns|gridRows, gridAutoRepeatColumns|gridAutoRepeatRows etc. to be
inline with the style specified. So does the getter.

Patch by Ziran Sun <Ziran Sun> on 2021-05-27
Reviewed by Antoine Quint.

This change has improved the two exising animation related grid-template-* tests.

• animation/CSSPropertyAnimation.cpp:

(WebCore::CSSPropertyAnimationWrapperMap::CSSPropertyAnimationWrapperMap):

12:14 PM Changeset in webkit [278172] by timothy_horton@apple.com

• 7 edits
  3 adds in trunk

Introduce BifurcatedGraphicsContext
​https://bugs.webkit.org/show_bug.cgi?id=226318

Reviewed by Sam Weinig.

Source/WebCore:

New test: BifurcatedGraphicsContextTests.BasicBifurcatedContext

Introduce a GraphicsContext subclass that forwards painting
to two arbitrary child GraphicsContexts. This can be used, for example,
to paint into a bitmap and display list simultaneously.

• Headers.cmake:
• Sources.txt:
• WebCore.xcodeproj/project.pbxproj:
• platform/graphics/BifurcatedGraphicsContext.cpp: Added.

(WebCore::BifurcatedGraphicsContext::BifurcatedGraphicsContext):
(WebCore::BifurcatedGraphicsContext::~BifurcatedGraphicsContext):
(WebCore::BifurcatedGraphicsContext::hasPlatformContext const):
(WebCore::BifurcatedGraphicsContext::platformContext const):
(WebCore::BifurcatedGraphicsContext::save):
(WebCore::BifurcatedGraphicsContext::restore):
(WebCore::BifurcatedGraphicsContext::drawRect):
(WebCore::BifurcatedGraphicsContext::drawLine):
(WebCore::BifurcatedGraphicsContext::drawEllipse):
(WebCore::BifurcatedGraphicsContext::applyStrokePattern):
(WebCore::BifurcatedGraphicsContext::applyFillPattern):
(WebCore::BifurcatedGraphicsContext::drawPath):
(WebCore::BifurcatedGraphicsContext::fillPath):
(WebCore::BifurcatedGraphicsContext::strokePath):
(WebCore::BifurcatedGraphicsContext::beginTransparencyLayer):
(WebCore::BifurcatedGraphicsContext::endTransparencyLayer):
(WebCore::BifurcatedGraphicsContext::applyDeviceScaleFactor):
(WebCore::BifurcatedGraphicsContext::fillRect):
(WebCore::BifurcatedGraphicsContext::fillRoundedRectImpl):
(WebCore::BifurcatedGraphicsContext::fillRectWithRoundedHole):
(WebCore::BifurcatedGraphicsContext::clearRect):
(WebCore::BifurcatedGraphicsContext::strokeRect):
(WebCore::BifurcatedGraphicsContext::fillEllipse):
(WebCore::BifurcatedGraphicsContext::strokeEllipse):
(WebCore::BifurcatedGraphicsContext::setIsCALayerContext):
(WebCore::BifurcatedGraphicsContext::isCALayerContext const):
(WebCore::BifurcatedGraphicsContext::setIsAcceleratedContext):
(WebCore::BifurcatedGraphicsContext::renderingMode const):
(WebCore::BifurcatedGraphicsContext::clip):
(WebCore::BifurcatedGraphicsContext::clipOut):
(WebCore::BifurcatedGraphicsContext::clipPath):
(WebCore::BifurcatedGraphicsContext::clipBounds const):
(WebCore::BifurcatedGraphicsContext::setLineCap):
(WebCore::BifurcatedGraphicsContext::setLineDash):
(WebCore::BifurcatedGraphicsContext::setLineJoin):
(WebCore::BifurcatedGraphicsContext::setMiterLimit):
(WebCore::BifurcatedGraphicsContext::drawNativeImage):
(WebCore::BifurcatedGraphicsContext::drawPattern):
(WebCore::BifurcatedGraphicsContext::scale):
(WebCore::BifurcatedGraphicsContext::rotate):
(WebCore::BifurcatedGraphicsContext::translate):
(WebCore::BifurcatedGraphicsContext::concatCTM):
(WebCore::BifurcatedGraphicsContext::setCTM):
(WebCore::BifurcatedGraphicsContext::getCTM const):
(WebCore::BifurcatedGraphicsContext::roundToDevicePixels):
(WebCore::BifurcatedGraphicsContext::drawFocusRing):
(WebCore::BifurcatedGraphicsContext::drawLinesForText):
(WebCore::BifurcatedGraphicsContext::drawDotsForDocumentMarker):
(WebCore::BifurcatedGraphicsContext::setURLForRect):
(WebCore::BifurcatedGraphicsContext::setDestinationForRect):
(WebCore::BifurcatedGraphicsContext::addDestinationAtPoint):
(WebCore::BifurcatedGraphicsContext::supportsInternalLinks const):
(WebCore::BifurcatedGraphicsContext::updateState):
(WebCore::BifurcatedGraphicsContext::deprecatedPrivateContext const):

• platform/graphics/BifurcatedGraphicsContext.h: Added.

Source/WebKit:

• Shared/RemoteLayerTree/CGDisplayListImageBufferBackend.cpp:

(WebKit::CGDisplayListImageBufferBackend::create):

Tools:

• TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
• TestWebKitAPI/Tests/WebCore/cg/BifurcatedGraphicsContextTestsCG.cpp: Added.

(TestWebKitAPI::TEST):

11:45 AM Changeset in webkit [278171] by Adrian Perez de Castro

• 1 copy in releases/WPE WebKit/webkit-2.33.1

WPE WebKit 2.33.1

11:45 AM Changeset in webkit [278170] by Adrian Perez de Castro

• 4 edits in trunk

Unreviewed. Update OptionsWPE.cmake and NEWS for the 2.33.1 release

.:

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

Source/WebKit:

• wpe/NEWS: Add release notes for 2.33.1

10:56 AM Changeset in webkit [278169] by Peng Liu

• 4 edits in trunk

Calling -[WKWebView _closeAllMediaPresentations] crashes
​https://bugs.webkit.org/show_bug.cgi?id=226305

Reviewed by Chris Dumez.

Source/WebKit:

Provide an empty completion handler to fix a crash.

• UIProcess/API/Cocoa/WKWebView.mm:

(-[WKWebView _closeAllMediaPresentations]):

Tools:

• TestWebKitAPI/Tests/WebKitCocoa/WKWebViewCloseAllMediaPresentations.mm:

(TEST):

10:54 AM Changeset in webkit [278168] by Devin Rousso

• 7 edits in trunk

Use NSLocale SPI for comparing language identifiers instead of string comparison
​https://bugs.webkit.org/show_bug.cgi?id=226252

Reviewed by Wenson Hsieh.

Source/WTF:

The existing indexOfBestMatchingLanguageInList logic uses (sub)string comparison of BCP47
language tags (e.g. looking for a '-' and comparing the substring before and after in the
input language and each item in the language list). While this does work for some languages,
it's really not the correct way to compare BCP47 language tags. Additionally, it basically
ignores the text after the first '-', unless there's an exact match with the input
language. This leads to "zh-CN" and "zh-HK" being considered a (not exact) match, which
isn't ideal.

• wtf/Language.cpp:
• wtf/cocoa/LanguageCocoa.mm:

(WTF::indexOfBestMatchingLanguageInList): Added.

• wtf/spi/cocoa/NSLocaleSPI.h:

LayoutTests:

• media/modern-media-controls/tracks-support/sorted-by-user-preferred-languages.html:
• media/modern-media-controls/tracks-support/sorted-by-user-preferred-languages-expected.txt:

10:42 AM Changeset in webkit [278167] by Simon Fraser

• 3 edits in trunk/Source/WebCore

Add animations to GraphicsLayer logging output
​https://bugs.webkit.org/show_bug.cgi?id=226303

Reviewed by Tim Horton.

Log animations in GraphicsLayer, so that it shows up in the "Layers" log channel,
and in showGraphicsLayerTree() output.

• platform/graphics/ca/GraphicsLayerCA.cpp:

(WebCore::operator<<):
(WebCore::GraphicsLayerCA::dumpAnimations):
(WebCore::GraphicsLayerCA::dumpAdditionalProperties const):

• platform/graphics/ca/GraphicsLayerCA.h:

10:07 AM Changeset in webkit [278166] by Brent Fulgham

• 2 edits in trunk/Source/WebCore

Pass frame ref during navigationPolicyCompletionHandler
​https://bugs.webkit.org/show_bug.cgi?id=226268
<rdar://problem/78512383>

Reviewed by Ryosuke Niwa.

Pass the active frame to the navigationPolicyCompletionHandler lambda as a Ref to
ensure it is kept alive during the policy check and subsequent steps.

• loader/DocumentLoader.cpp:

(WebCore::DocumentLoader::willSendRequest):

10:07 AM Changeset in webkit [278165] by commit-queue@webkit.org

• 3 edits in trunk/Source/WebKit

[GTK4] Navigation swipe snapshot always times out
​https://bugs.webkit.org/show_bug.cgi?id=226332

Patch by Alexander Mikhaylenko <Alexander Mikhaylenko> on 2021-05-27
Reviewed by Michael Catanzaro.

Enable ViewGestureGeometryCollector for GTK4 as well.

• WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::close):
(WebKit::WebPage::mainFrameDidLayout):

• WebProcess/WebPage/WebPage.h:

6:27 AM Changeset in webkit [278164] by Alan Bujtas

• 4 edits
  2 adds in trunk

REGRESSION (r244434) Dispatch intrinsicContentsSizeChanged on the main frame only
​https://bugs.webkit.org/show_bug.cgi?id=226311

Reviewed by Tim Horton.

Source/WebCore:

While the Mail version of the content autosizing runs on both the main frame and the subframe(s),
the content size reporting should only happen on the main frame. The client is only interested
in the enclosing size of the content (also the API to retrieve the information is per-WKWebView,
it doesn't expose this information on a frame-by-frame basis).

Test: fast/dynamic/mail-autosize-with-iframe.html

• page/FrameView.cpp:

(WebCore::FrameView::autoSizeIfEnabled):
(WebCore::FrameView::enableFixedWidthAutoSizeMode):
(WebCore::FrameView::enableSizeToContentAutoSizeMode):
(WebCore::FrameView::enableAutoSizeMode):

LayoutTests:

• fast/dynamic/mail-autosize-with-iframe-expected.txt: Added.
• fast/dynamic/mail-autosize-with-iframe.html: Added.

6:26 AM Changeset in webkit [278163] by weinig@apple.com

• 7 edits
  2 moves
  1 add
  4 deletes in trunk/Source/WebCore

Merge WorkerGlobalScope+Caches.idl and DOMWindow+Caches.idl into a single WindowOrWorkerGlobalScope+Caches.idl
​https://bugs.webkit.org/show_bug.cgi?id=226271

Reviewed by Darin Adler.

Address FIXMEs in both WorkerGlobalScope+Caches.idl and DOMWindow+Caches.idl by
merging them into a single WindowOrWorkerGlobalScope+Caches.idl, matching the spec
langauge more closely.

• CMakeLists.txt:
• DerivedSources-input.xcfilelist:
• DerivedSources-output.xcfilelist:
• DerivedSources.make:
• Modules/cache/DOMWindow+Caches.idl: Removed.
• Modules/cache/DOMWindowCaches.cpp: Removed.
• Modules/cache/DOMWindowCaches.h: Removed.
• Modules/cache/WindowOrWorkerGlobalScope+Caches.idl: Added.
• Modules/cache/WindowOrWorkerGlobalScopeCaches.cpp: Added.

(WebCore::DOMWindowCaches::supplementName):
(WebCore::WorkerGlobalScopeCaches::supplementName):
(WebCore::DOMWindowCaches::DOMWindowCaches):
(WebCore::DOMWindowCaches::from):
(WebCore::DOMWindowCaches::caches const):
(WebCore::WorkerGlobalScopeCaches::WorkerGlobalScopeCaches):
(WebCore::WorkerGlobalScopeCaches::from):
(WebCore::WorkerGlobalScopeCaches::caches const):
(WebCore::WindowOrWorkerGlobalScopeCaches::caches):

• Modules/cache/WindowOrWorkerGlobalScopeCaches.h: Added.
• Modules/cache/WorkerGlobalScope+Caches.idl: Removed.
• Modules/cache/WorkerGlobalScopeCaches.cpp: Removed.
• Modules/cache/WorkerGlobalScopeCaches.h: Removed.
• Sources.txt:
• WebCore.xcodeproj/project.pbxproj:

5:30 AM Changeset in webkit [278162] by Adrian Perez de Castro

• 11 edits in trunk/Source

Non-unified build fixes, late May 2021 edition
​https://bugs.webkit.org/show_bug.cgi?id=226322

Unreviewed non-unified build fixes.

Source/JavaScriptCore:

• jit/JITPlan.cpp: Add missing HeapInlines.h and VMInlines.h headers.
• jit/JITWorklist.cpp: Add missing HeapInlines.h and SlotVisitorInlines.h headers,

remove SlotVisitor.h which is now unneeded.

• jit/JITWorklistThread.cpp: Add missing HeapInlines.h header.

Source/WebCore:

No new tests needed.

• css/CSSGridIntegerRepeatValue.cpp: Add missing wtf/text/StringConcatenateNumbers.h

header.

• css/CSSTimingFunctionValue.cpp: Ditto.
• layout/formattingContexts/flex/FlexFormattingContext.h: Add missing FormattingQuirks.h

header, remove now unneeded FormattingContext.h header.

• loader/PrivateClickMeasurement.cpp: Add missing wtf/text/StringConcatenateNumbers.h

header.

• platform/graphics/cairo/GraphicsContextCairo.cpp: Add missing Gradient.h header.
• platform/graphics/filters/FEMorphology.cpp: Add missing ColorTypes.h header.

5:22 AM Changeset in webkit [278161] by Wenson Hsieh

• 7 edits in trunk/Source/WebKit

[GPU Process] Add validation when reading DisplayListSharedMemoryHeader::waitingStatus in DisplayListReaderHandle
​https://bugs.webkit.org/show_bug.cgi?id=226306
rdar://78118900

Reviewed by Chris Dumez.

Store this waiting status (an enum class with three valid states) as a raw uint8_t in shared memory, rather
than an enum class. This allows us to verify that this value indicates a valid waiting status (using
isValidEnum) upon reading this flag in the GPU process, and perform a MESSAGE_CHECK in the case where it is
invalid.

• GPUProcess/graphics/DisplayListReaderHandle.h:

(WebKit::DisplayListReaderHandle::startWaiting):
(WebKit::DisplayListReaderHandle::stopWaiting):

• GPUProcess/graphics/RemoteRenderingBackend.cpp:

(WebKit::RemoteRenderingBackend::nextDestinationImageBufferAfterApplyingDisplayLists):

• Shared/SessionState.cpp:

(WebKit::PageState::decode):

• Shared/SessionState.h:

Drive-by fix: remove a non-templated declaration of isValidEnum(). This is necessary in order to use
WTF::isValidEnum without the explicit WTF namespace prefix inside DisplayListReaderHandle, since WTF's
declaration of isValidEnum clashes with this declaration.

Instead of using this separate method for validating the enum, simply encode and decode the
shouldOpenExternalURLsPolicy enum using << and >> operators, respectively; this ensures that the decoded value
of shouldOpenExternalURLsPolicy is either valid, or the result will be WTF::nullopt if decoding fails or
results in an invalid enum.

• Shared/SharedDisplayListHandle.h:
• WebProcess/GPU/graphics/DisplayListWriterHandle.h:

(WebKit::DisplayListWriterHandle::tryToResume):

5:19 AM Changeset in webkit [278160] by Adrian Perez de Castro

• 2 edits in trunk

Unreviewed. [WPE] Bump version numbers

• Source/cmake/OptionsWPE.cmake:

4:49 AM Changeset in webkit [278159] by Angelos Oikonomopoulos

• 2 edits in trunk/Tools

[JSC] Implement high-level retry loop for run-jsc-stress-tests
​https://bugs.webkit.org/show_bug.cgi?id=225803

Reviewed by Adrian Perez de Castro.

Now that we have information on which tests produced a result and
which tests were never heard of, introduce retry logic for the
incomplete tests. This should only modify the behavior when using
--gnu-parallel-runner.

The issue here is that the MIPS boards currently in use for JSC
testing routinely crash and, often enough, end up with filesystem
corruption and need to have their R/W storage reinitialized.

Work around this by re-executing the tests that we weren't able to get
a result for. When some tests failed to complete, go through the
process of remote initilization (copying over and unpacking the
bundle) from scratch, so that remotes that had their remoteDirectory
wiped will get back into play.

To make this more efficient, introduce in-band signaling to let
run-jsc-stress-tests know that a remote lost its state. In response,
we kill the GNU parallel process and re-initialize all boards. This
helps for buildbot workers that only have a few remote boards assigned
and apparently also works around a bug in GNU parallel 20161222
(currently the most widely deployed version) which causes it to run
out of open files and get stuck in a busy loop.

While here, make run-jsc-stress-tests more robust against corrupted
status files, i.e. ignore empty ones (observed in practice) and use a
regex to verify the contents of non-empty state files.

For consistency, use the same logic for parsing status files when
doing local runs.

In testing with two remotes, run-jsc-stress-tests was able to recover
from either one or both of its remotes rebooting and wiping their
local storage in the middle of a test run.

• Scripts/run-jsc-stress-tests:

3:58 AM WebKitGTK/2.32.x edited by clopez@igalia.com

(diff)

3:19 AM Changeset in webkit [278158] by youenn@apple.com

• 7 edits
  2 adds in trunk

Support H264 profiles in MediaRecorder
​https://bugs.webkit.org/show_bug.cgi?id=226219
<rdar://78027944>

Reviewed by Eric Carlson.

Source/WebCore:

Use mimeType option given in MediaRecorder options to detect which H264 profile to use.
Implement this by parsing the mime type codec parameter and use the avc1.WXYZ string to compute the profile.
By default, use baseline profile, which is not VideoToolbox profile but has wider decoding support.
If profile specified by the application fails, we downgrade to baseline.

Test: http/wpt/mediarecorder/MediaRecorder-video-h264-profiles.html

• Modules/mediarecorder/MediaRecorderProvider.cpp:

(WebCore::MediaRecorderProvider::isSupported):

• platform/mediarecorder/cocoa/MediaRecorderPrivateWriterCocoa.h:
• platform/mediarecorder/cocoa/MediaRecorderPrivateWriterCocoa.mm:

(WebCore::MediaRecorderPrivateWriter::create):
(WebCore::MediaRecorderPrivateWriter::initialize):
(WebCore::MediaRecorderPrivateWriter::setOptions): Deleted.

• platform/mediarecorder/cocoa/VideoSampleBufferCompressor.h:
• platform/mediarecorder/cocoa/VideoSampleBufferCompressor.mm:

(WebCore::VideoSampleBufferCompressor::create):
(WebCore::VideoSampleBufferCompressor::VideoSampleBufferCompressor):
(WebCore::VideoSampleBufferCompressor::vtProfileLevel const):
(WebCore::VideoSampleBufferCompressor::initCompressionSession):

LayoutTests:

• http/wpt/mediarecorder/MediaRecorder-video-h264-profiles-expected.txt: Added.
• http/wpt/mediarecorder/MediaRecorder-video-h264-profiles.html: Added.

3:01 AM WebKitGTK/2.32.x edited by Adrian Perez de Castro

(diff)

2:56 AM Changeset in webkit [278157] by commit-queue@webkit.org

• 2 edits in trunk/Source/JavaScriptCore

[JSC] Fix crash on 32-bit big endian systems.
​https://bugs.webkit.org/show_bug.cgi?id=226264

Patch by Daniel Kolesa <Daniel Kolesa> on 2021-05-27
Reviewed by Caio Araujo Neponoceno de Lima.

This is an instance where properly offsetting was missed since
the issue was not present in 2.30 series and therefore not fixed
by r273104.

• llint/LowLevelInterpreter32_64.asm:

2:51 AM WebKitGTK/2.32.x edited by Adrian Perez de Castro

(diff)

2:43 AM WebKitGTK/2.32.x edited by Adrian Perez de Castro

(diff)

2:36 AM Changeset in webkit [278156] by commit-queue@webkit.org

• 2 edits in trunk/Source/WebKit

Transient quarter display with a HiDPI /4k screen and a 200% scaling
​https://bugs.webkit.org/show_bug.cgi?id=219202

Patch by Alexander Mikhaylenko <Alexander Mikhaylenko> on 2021-05-27
Reviewed by Adrian Perez de Castro.

Set the root layer transformation before syncing animations and not after.
This way we avoid having the first frame use the wrong scale on hidpi.

• Shared/CoordinatedGraphics/CoordinatedGraphicsScene.cpp:

(WebKit::CoordinatedGraphicsScene::paintToCurrentGLContext):

2:18 AM Changeset in webkit [278155] by commit-queue@webkit.org

• 6 edits in trunk/LayoutTests

MacOS WebM Format Reader returns enabled for tracks that do not have samples
​https://bugs.webkit.org/show_bug.cgi?id=226078
<rdar://74048072>

Patch by Kimmo Kinnunen <​kkinnunen@apple.com> on 2021-05-27
Reviewed by Jer Noble.

After r274378 the WebM format reader does not support WebM with non-disabled audio tracks
that do not have any samples.
Remove the empty audio track from all red-green.webmvp9.webm files so that WebGL
conformance tests that use the file do not fail due to this.

• fast/canvas/webgl/resources/red-green.webmvp8.webm:
• http/tests/webgl/1.0.2/resources/webgl_test_files/conformance/resources/red-green.webmvp8.webm:
• webgl/1.0.3/resources/webgl_test_files/conformance/resources/red-green.webmvp8.webm:
• webgl/2.0.0/resources/webgl_test_files/resources/red-green.webmvp8.webm:
• webgl/resources/webgl_test_files/resources/red-green.webmvp8.webm:

2:07 AM Changeset in webkit [278154] by Carlos Garcia Campos

• 2 edits in trunk/Source/WebCore

[SOUP] Do not disable the idle connection timeout
​https://bugs.webkit.org/show_bug.cgi?id=226263

Reviewed by Sergio Villar Senin.

We disable the connection timeout in SoupSession because we implement our own timeout in WebKit, but that's not
the case for the idle connection timeout. There's no reason keep idle connection alive forever.

• platform/network/soup/SoupNetworkSession.cpp:

(WebCore::SoupNetworkSession::SoupNetworkSession):

1:17 AM Changeset in webkit [278153] by commit-queue@webkit.org

• 3 edits in trunk/Source/ThirdParty/ANGLE

ANGLE could be compiled with "aggressive uninitialised variable" setting
​https://bugs.webkit.org/show_bug.cgi?id=226260

Patch by Kimmo Kinnunen <​kkinnunen@apple.com> on 2021-05-27
Reviewed by Alexey Proskuryakov.

Turn on compiler flag to warn about uninitialized variables that might be
conditionally uninitialized. This is the same flag as upstream ANGLE uses.
Fix a spurious warning produced by this flag.
This helps in scenarios where custom targets compile ANGLE sources and
this flag set, such as tests.

• Configurations/Base.xcconfig:
• src/compiler/translator/TranslatorMetalDirect/FixTypeConstructors.cpp:

1:14 AM Changeset in webkit [278152] by commit-queue@webkit.org

• 7 edits in trunk/Source/ThirdParty/ANGLE

Cherry-pick ANGLE: D3D11: Skip blits if there is no intersection of dest areas
​https://bugs.webkit.org/show_bug.cgi?id=225190
<rdar://77084155>

Patch by Kimmo Kinnunen <​kkinnunen@apple.com> on 2021-05-27
Reviewed by David Kilzer.

Cherry-pick ANGLE commit b574643ef28c92fcea5122dd7a72acb42a514eed
Fixes a security issue on D3D11.
Potential a correctness issue on some OpenGL drivers.
No effect on Metal, but the nodiscard part is still useful.

Upstream description:
D3D11: Skip blits if there is no intersection of dest areas

Blit11 would clip the destination rectangle with the destination size
but ignore the result. gl::ClipRectangle returns false when the
rectangles do not intersect at all, indicating the blit can be skipped.

This could lead to an out-of-bounds write to the GPU memory for the
destination texture.

Mark ClipRectangle as nodiscard to prevent future issues.

• src/libANGLE/angletypes.h:
• src/libANGLE/renderer/d3d/d3d11/Blit11.cpp:
• src/libANGLE/renderer/gl/FramebufferGL.cpp:

(rx::FramebufferGL::clipSrcRegion):

• src/libANGLE/renderer/metal/ContextMtl.mm:

(rx::ContextMtl::updateScissor):

• src/libANGLE/renderer/vulkan/ContextVk.cpp:

(rx::ContextVk::updateScissor):

• src/tests/gl_tests/BlitFramebufferANGLETest.cpp:

(TEST_P):