Timeline
Jun 15, 2021:
- 10:40 PM Changeset in webkit [278922] by
-
- 2 edits in trunk/Source/WebCore
some http/tests/ are crashing with ASSERTION FAILED: isInRoutingArbitrationForToken(token) m_setupArbitrationOngoing https://bugs.webkit.org/show_bug.cgi?id=226896
Reviewed by Eric Carlson.
When the
RoutingArbiterfails to-[beginArbitrationWithCategory:completionHandler:],
a token won't be added tom_tokens. Therefore, we should remove the assertion
inSharedRoutingArbitrator::endRoutingArbitrationForToken().
- platform/audio/mac/SharedRoutingArbitrator.mm:
(WebCore::SharedRoutingArbitrator::endRoutingArbitrationForToken):
- 9:55 PM Changeset in webkit [278921] by
-
- 7 edits1 add in trunk
[JSC] Optimize JSON.parse with small content by dropping single character Identifier pool
https://bugs.webkit.org/show_bug.cgi?id=227057
Reviewed by Sam Weinig.
JSTests:
- microbenchmarks/flight-todomvc-json.js: Added.
(test):
Source/JavaScriptCore:
Profiler results and investigation suggest interesting things.
- Sampling profiler says Flight-TodoMVC is mostly JSON.parse benchmark.
- Each JSON data of Flight-TodoMVC is small. And JSON.parse is super frequently called.
- In the case of JSON.parse with small data, LiteralParser's construction / destruction costs much since it has large Identifier pool with std::array<>.
As a simple first step, this patch removes single character Identifier pool from LiteralParser since
the exact same Identifier data can be retrieved from VM's SmallStrings.
We created a microbenchmark from Flight-TodoMVC's data, and the result is roughly 20% better.
And we observed 0.6% improvement in Speedometer2.
ToT Patched
flight-todomvc-json 81.0552+-0.8403 67.5756+-0.6221 definitely 1.1995x faster
----------------------------------------------------------------------------------------------------------------------------------
| subtest | ms | ms | b / a | pValue (significance using False Discovery Rate) |
----------------------------------------------------------------------------------------------------------------------------------
| Elm-TodoMVC |128.991667 |128.450000 |0.995801 | 0.278228 |
| VueJS-TodoMVC |28.487500 |27.925000 |0.980254 | 0.139315 |
| EmberJS-TodoMVC |133.950000 |134.175000 |1.001680 | 0.685021 |
| BackboneJS-TodoMVC |51.670833 |51.537500 |0.997420 | 0.628993 |
| Preact-TodoMVC |21.783333 |21.754167 |0.998661 | 0.944237 |
| AngularJS-TodoMVC |143.820833 |143.770833 |0.999652 | 0.933953 |
| Vanilla-ES2015-TodoMVC |71.608333 |71.416667 |0.997323 | 0.500591 |
| Inferno-TodoMVC |69.179167 |69.525000 |1.004999 | 0.412406 |
| Flight-TodoMVC |81.354167 |79.020833 |0.971319 | 0.000053 (significant) |
| Angular2-TypeScript-TodoMVC |42.654167 |41.887500 |0.982026 | 0.086053 |
| VanillaJS-TodoMVC |57.054167 |56.633333 |0.992624 | 0.176804 |
| jQuery-TodoMVC |274.595833 |275.670833 |1.003915 | 0.148812 |
| EmberJS-Debug-TodoMVC |358.387500 |357.595833 |0.997791 | 0.323387 |
| React-TodoMVC |93.804167 |93.329167 |0.994936 | 0.113410 |
| React-Redux-TodoMVC |157.954167 |157.266667 |0.995647 | 0.131298 |
| Vanilla-ES2015-Babel-Webpack-TodoMVC |68.687500 |68.054167 |0.990779 | 0.002155 (significant) |
----------------------------------------------------------------------------------------------------------------------------------
a mean = 235.28964
b mean = 236.72163
pValue = 0.0121265559
(Bigger means are better.)
1.006 times better
Results ARE significant
- runtime/Identifier.h:
(JSC::Identifier::canUseSingleCharacterString):
- runtime/LiteralParser.cpp:
(JSC::LiteralParser<CharType>::makeIdentifier):
- runtime/LiteralParser.h:
- runtime/SmallStrings.cpp:
(JSC::SmallStrings::singleCharacterStringRep):
- runtime/SmallStrings.h:
- 9:25 PM Changeset in webkit [278920] by
-
- 10 edits1 delete in trunk/Source/WebKit
Remove unused code on API::HTTPCookieStore
https://bugs.webkit.org/show_bug.cgi?id=227056
Reviewed by Brady Eidson.
The dust has settled a bit since r267763.
Let's clean up!
- SourcesCocoa.txt:
- UIProcess/API/APIHTTPCookieStore.cpp:
(API::HTTPCookieStore::unregisterObserver):
(API::HTTPCookieStore::cookieManagerDestroyed):
(API::HTTPCookieStore::flushDefaultUIProcessCookieStore): Deleted.
(API::HTTPCookieStore::getAllDefaultUIProcessCookieStoreCookies): Deleted.
(API::HTTPCookieStore::setCookieInDefaultUIProcessCookieStore): Deleted.
(API::HTTPCookieStore::deleteCookieFromDefaultUIProcessCookieStore): Deleted.
(API::HTTPCookieStore::startObservingChangesToDefaultUIProcessCookieStore): Deleted.
(API::HTTPCookieStore::stopObservingChangesToDefaultUIProcessCookieStore): Deleted.
(API::HTTPCookieStore::deleteCookiesInDefaultUIProcessCookieStore): Deleted.
(API::HTTPCookieStore::setHTTPCookieAcceptPolicyInDefaultUIProcessCookieStore): Deleted.
- UIProcess/API/APIHTTPCookieStore.h:
- UIProcess/API/Cocoa/APIHTTPCookieStoreCocoa.mm: Removed.
- UIProcess/API/Cocoa/WKBrowsingContextController.mm:
- UIProcess/API/Cocoa/WKDownloadInternal.h:
- UIProcess/Cocoa/WebProcessProxyCocoa.mm:
(WebKit::WebProcessProxy::sendAudioComponentRegistrations):
- UIProcess/Cocoa/WebViewImpl.mm:
(WebKit::WebViewImpl::handleClickForDataDetectionResult):
- UIProcess/WebProcessPool.h:
- WebKit.xcodeproj/project.pbxproj:
- 8:58 PM Changeset in webkit [278919] by
-
- 6 edits in trunk/Source/WebCore
Cleanup RenderElement::RendererCreationType
https://bugs.webkit.org/show_bug.cgi?id=227033
Reviewed by Simon Fraser.
Let the caller define the preferred type of renderer for certain display type values.
- html/HTMLFieldSetElement.cpp:
(WebCore::HTMLFieldSetElement::createElementRenderer):
- html/HTMLSummaryElement.cpp:
(WebCore::HTMLSummaryElement::createElementRenderer):
- rendering/RenderElement.cpp:
(WebCore::RenderElement::createFor):
- rendering/RenderElement.h:
(WebCore::RenderElement::createFor):
- 6:47 PM Changeset in webkit [278918] by
-
- 10 edits2 adds in trunk
[GPU Process] ConcreteImageBuffer::baseTransform() returns incorrect transform for unaccelerated ImageBuffer
https://bugs.webkit.org/show_bug.cgi?id=227044
<rdar://78642742>
Reviewed by Simon Fraser.
Source/WebCore:
The static constant ImageBufferBackend::isOriginAtUpperLeftCorner is used
to tell where the origin of the physical graphics context. For CG platforms,
it is at the bottom left corner. This requires the coordinates system to
be moved to the top-left corner and flipped.
The fix is:
1) Rename isOriginAtUpperLeftCorner to isOriginAtBottomLeftCorner since
it is a description for the physical graphics context. All the logical
graphics contexts have to have their coordinates system at the top-left
corner.
2) Set isOriginAtBottomLeftCorner to true in ImageBufferCGBackend only
so all the super classes inherit the 'true' value. This includes
ImageBufferShareableBitmapBackend for CG platforms.
Test: fast/canvas/canvas-large-dimensions-drawing.html
- platform/graphics/ConcreteImageBuffer.h:
- platform/graphics/ImageBufferBackend.h:
- platform/graphics/cg/ImageBufferCGBackend.h:
- platform/graphics/cg/ImageBufferCGBitmapBackend.h:
- platform/graphics/cg/ImageBufferIOSurfaceBackend.h:
Source/WebKit:
- Shared/RemoteLayerTree/CGDisplayListImageBufferBackend.h:
- WebProcess/GPU/graphics/cocoa/ImageBufferShareableIOSurfaceBackend.h:
LayoutTests:
The large size of the canvas in this layout test will force creating
unaccelerated ImageBufferBackend.
- fast/canvas/canvas-large-dimensions-drawing-expected.html: Added.
- fast/canvas/canvas-large-dimensions-drawing.html: Added.
- 6:01 PM Changeset in webkit [278917] by
-
- 7 edits2 adds in trunk
REGRESSION: MSE: Netflix playback sometimes fails when going to "next episode" in series
https://bugs.webkit.org/show_bug.cgi?id=226952
rdar://77583769
Reviewed by Eric Carlson.
Source/WebKit:
Following bug #225396, the remote/proxy MediaPlayer readyState became decoupled
from the currentTime. In the scenario where a time update was pushed from the
GPU to the WebContent process, followed by a change of readyState status; should
the HTMLMediaElement query the MediaPlayer's time in-between those two IPC calls,
it would be possible for the time to be incorrectly estimated as it assumed it
had progressed since it received the update.
Additionally, it was assumed that the only case where the time won't progress
was if the media element was explicitly paused. This was incorrect it could
have stalled due to the lack of data such as commonly happen with MSE.
We add in the time update IPC message a boolean indicating if the GPU's
MediaPlayer is either stalled or paused and that the time shouldn't advance.
While this new boolean appears redundant with the values of the cached
readyState/paused it differs in that timeIsProgressing is accurate at the
moment the time update is sent and received.
Test: media/media-source/media-source-seek-unbuffered.html
- GPUProcess/media/RemoteMediaPlayerProxy.cpp:
(WebKit::RemoteMediaPlayerProxy::mediaPlayerPausedOrStalled const): Add utility method.
(WebKit::RemoteMediaPlayerProxy::currentTimeChanged):
- GPUProcess/media/RemoteMediaPlayerProxy.h:
- WebProcess/GPU/media/MediaPlayerPrivateRemote.cpp:
(WebKit::MediaPlayerPrivateRemote::currentMediaTime const):
(WebKit::MediaPlayerPrivateRemote::seek): Set the cached media time to the seek
value so that reading it back the time will provide accurate data.
(WebKit::MediaPlayerPrivateRemote::seekWithTolerance): same as with seek method.
(WebKit::MediaPlayerPrivateRemote::currentTimeChanged): Check if timeIsProgressing
is true and only estimate the time if we are certain playback has continued.
- WebProcess/GPU/media/MediaPlayerPrivateRemote.h:
- WebProcess/GPU/media/MediaPlayerPrivateRemote.messages.in: add timeIsProgressing
to method.
LayoutTests:
- media/media-source/media-source-seek-unbuffered-expected.txt: Added.
- media/media-source/media-source-seek-unbuffered.html: Added.
- 5:56 PM Changeset in webkit [278916] by
-
- 3 edits in trunk/Source/WebKit
Use more SQL transactions in ResourceLoadStatisticsDatabaseStore
https://bugs.webkit.org/show_bug.cgi?id=227034
Reviewed by Kate Cheney.
Use more SQL transactions in ResourceLoadStatisticsDatabaseStore, for performance.
I updated the code so that we start a SQL transaction whenever we do more than one
consecutive "write" statement.
- NetworkProcess/Classifier/ResourceLoadStatisticsDatabaseStore.cpp:
(WebKit::ResourceLoadStatisticsDatabaseStore::addMissingColumnsToTable):
(WebKit::ResourceLoadStatisticsDatabaseStore::addMissingTablesIfNecessary):
(WebKit::ResourceLoadStatisticsDatabaseStore::populateFromMemoryStore):
(WebKit::ResourceLoadStatisticsDatabaseStore::mergeStatistics):
(WebKit::ResourceLoadStatisticsDatabaseStore::reclassifyResources):
(WebKit::ResourceLoadStatisticsDatabaseStore::grandfatherDataForDomains):
(WebKit::ResourceLoadStatisticsDatabaseStore::insertExpiredStatisticForTesting):
- 5:46 PM Changeset in webkit [278915] by
-
- 1 copy in tags/Safari-612.1.18.1.2
Tag Safari-612.1.18.1.2.
- 5:00 PM Changeset in webkit [278914] by
-
- 1 edit13 adds in trunk/WebKitLibraries
[tvOS 15] Support building WebKit
https://bugs.webkit.org/show_bug.cgi?id=227037
<rdar://problem/79354143>
Reviewed by Alex Christensen.
- WebKitPrivateFrameworkStubs/appletvos/15/AppSupport.framework/AppSupport.tbd: Added.
- WebKitPrivateFrameworkStubs/appletvos/15/CorePrediction.framework/CorePrediction.tbd: Added.
- WebKitPrivateFrameworkStubs/appletvos/15/FileProvider.framework/FileProvider.tbd: Added.
- WebKitPrivateFrameworkStubs/appletvos/15/GraphicsServices.framework/GraphicsServices.tbd: Added.
- WebKitPrivateFrameworkStubs/appletvos/15/IOSurfaceAccelerator.framework/IOSurfaceAccelerator.tbd: Added.
- WebKitPrivateFrameworkStubs/appletvos/15/RunningBoardServices.framework/RunningBoardServices.tbd: Added.
- 4:49 PM Changeset in webkit [278913] by
-
- 18 edits in trunk/Source/WebKit
Reloading the view should be able to recover if the GPUProcess or NetworkProcess are hung
https://bugs.webkit.org/show_bug.cgi?id=227051
Reviewed by Geoffrey Garen.
Reloading the view should be able to get us out of bad state if the GPUProcess or NetworkProcess are hung.
This is useful as the first instinct of the user may be to reload the page if the page appears hung.
Before the change, if the page is in a bad state due to a hung GPUProcess or NetworkProcess and you'd
reload, the reload would hang indefinitely. After the change, we'll make sure that both processes
are still responsive when triggering the reload. If they are unresponsive, we kill them and the reload
is able to proceed.
- GPUProcess/GPUProcess.cpp:
(WebKit::GPUProcess::didReceiveMessage):
- GPUProcess/GPUProcess.h:
- GPUProcess/GPUProcess.messages.in:
- Shared/AuxiliaryProcess.cpp:
(WebKit::AuxiliaryProcess::mainThreadPing):
- Shared/AuxiliaryProcess.h:
- Shared/AuxiliaryProcess.messages.in:
- UIProcess/AuxiliaryProcessProxy.cpp:
(WebKit::AuxiliaryProcessProxy::checkForResponsiveness):
- UIProcess/AuxiliaryProcessProxy.h:
- UIProcess/GPU/GPUProcessProxy.messages.in:
- UIProcess/Network/NetworkProcessProxy.messages.in:
- UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::reload):
- UIProcess/WebProcessProxy.cpp:
(WebKit::WebProcessProxy::isResponsive):
(WebKit::WebProcessProxy::isResponsiveWithLazyStop):
- UIProcess/WebProcessProxy.h:
- UIProcess/WebProcessProxy.messages.in:
- WebProcess/WebProcess.cpp:
- WebProcess/WebProcess.h:
- WebProcess/WebProcess.messages.in:
- 4:31 PM Changeset in webkit [278912] by
-
- 2 edits1 delete in trunk/LayoutTests
[GLIB] Unreviewed test gardening. Updated test expectations after r277896.
Two tests are now failing:
- fast/canvas/canvas-imageSmoothingEnabled.html.
- fast/canvas/canvas-imageSmoothingQuality.html.
- platform/glib/TestExpectations:
- platform/gtk/fast/canvas/canvas-imageSmoothingQuality-expected.txt: Removed.
- 4:28 PM Changeset in webkit [278911] by
-
- 1 edit2 adds in trunk/WebKitLibraries
[iOS 15] Support building WebKit (Part 2)
https://bugs.webkit.org/show_bug.cgi?id=227002
<rdar://problem/79316598>
Reviewed by Alex Christensen.
- WebKitPrivateFrameworkStubs/iOS/15/IOSurfaceAccelerator.framework/IOSurfaceAccelerator.tbd: Added.
- 4:12 PM Changeset in webkit [278910] by
-
- 3 edits19 adds in trunk
[watchOS 8] Support building WebKit
https://bugs.webkit.org/show_bug.cgi?id=227050
<rdar://problem/79365320>
Reviewed by Tim Horton.
Source/WebKit:
Covered by existing tests.
- Platform/spi/watchos/PepperUICoreSPI.h: Add QuickboardController SPI.
WebKitLibraries:
- WebKitPrivateFrameworkStubs/watchos/8/AppSupport.framework/AppSupport.tbd: Added.
- WebKitPrivateFrameworkStubs/watchos/8/CorePrediction.framework/CorePrediction.tbd: Added.
- WebKitPrivateFrameworkStubs/watchos/8/FileProvider.framework/FileProvider.tbd: Added.
- WebKitPrivateFrameworkStubs/watchos/8/GraphicsServices.framework/GraphicsServices.tbd: Added.
- WebKitPrivateFrameworkStubs/watchos/8/IOSurfaceAccelerator.framework/IOSurfaceAccelerator.tbd: Added.
- WebKitPrivateFrameworkStubs/watchos/8/PepperUICore.framework/PepperUICore.tbd: Added.
- WebKitPrivateFrameworkStubs/watchos/8/RunningBoardServices.framework/RunningBoardServices.tbd: Added.
- WebKitPrivateFrameworkStubs/watchos/8/SafariSafeBrowsing.framework/SafariSafeBrowsing.tbd: Added.
- WebKitPrivateFrameworkStubs/watchos/8/URLFormatting.framework/URLFormatting.tbd: Added.
- 3:50 PM Changeset in webkit [278909] by
-
- 2 edits in branches/safari-611-branch/Source/WebKit
Apply patch. rdar://problem/78875336
- 3:50 PM Changeset in webkit [278908] by
-
- 5 edits in branches/safari-611-branch/Source/WebKit
Cherry-pick r277536. rdar://problem/78875336
This reverts r278896.
- 3:06 PM Changeset in webkit [278907] by
-
- 9 edits in trunk/Source/WebCore
iOS - VoiceOver reads incorrectly in content editable element if role="document".
https://bugs.webkit.org/show_bug.cgi?id=227035
rdar://78776169
Reviewed by Chris Fleizach.
On iOS, VoiceOver was announcing <div contenteditable="true" role="document">
as a landmark and not as editable text. Furthermore, VoiceOver wasn't
echoing keyboard input once the element became interactive.
The solution is to return AccessibilityRole::TextArea for an element
with these attributes instead of Document.
Used shouldIgnoreRoleAttribute to special case this scenario.
- accessibility/AccessibilityObject.h:
(WebCore::AccessibilityObject::shouldIgnoreAttributeRole const):
- accessibility/AccessibilityObjectInterface.h:
Removed shouldIgnoreRoleAttribute from the AXObject interface since it
doesn't need to be exposed outside the AXObject class hierarchy.
- accessibility/AccessibilityRenderObject.cpp:
(WebCore::AccessibilityRenderObject::shouldIgnoreAttributeRole const):
- accessibility/AccessibilityRenderObject.h:
- accessibility/AccessibilityTreeItem.h:
- accessibility/ios/WebAccessibilityObjectWrapperIOS.mm:
(-[WebAccessibilityObjectWrapper _accessibilityTextEntryTraits]): Code cleanup.
- accessibility/isolatedtree/AXIsolatedObject.cpp:
(WebCore::AXIsolatedObject::shouldIgnoreAttributeRole const): Deleted.
- accessibility/isolatedtree/AXIsolatedObject.h:
- 3:00 PM Changeset in webkit [278906] by
-
- 2 edits in trunk/Source/ThirdParty/libwebrtc
Enable kVTVideoEncoderSpecification_RequiredLowLatency in case of MacOS software encoder
https://bugs.webkit.org/show_bug.cgi?id=226873
Reviewed by Eric Carlson.
In case software encoder is being created for baseline, we can use kVTVideoEncoderSpecification_RequiredLowLatency
which provides better bitrate management than the regular SW encoder while still generating baseline content.
We check whether SW encoder is created based on kVTCompressionPropertyKey_UsingHardwareAcceleratedVideoEncoder.
If property is unsupported or if its value is false, it means the encoder is software.
- Source/webrtc/sdk/objc/components/video_codec/RTCVideoEncoderH264.mm:
(-[RTCVideoEncoderH264 resetCompressionSessionWithPixelFormat:]):
- 2:49 PM Changeset in webkit [278905] by
-
- 9 edits in branches/safari-611-branch/Source/WebCore
Apply patch. rdar://problem/79355311
- 2:49 PM Changeset in webkit [278904] by
-
- 3 edits1 add in branches/safari-611-branch
Cherry-pick r278819. rdar://problem/79355258
https://bugs.webkit.org/show_bug.cgi?id=226576
<rdar://problem/78810362>
Reviewed by Yusuke Suzuki.
JSTests:
- stress/short-circuit-read-modify-write-cant-write-dst-before-tdz-check.js: Added. (let.result.eval.try.captureV): (catch):
Source/JavaScriptCore:
ShortCircuitReadModifyResolveNode can't emit a value into
its result until after it emits a TDZ check. We were temporarily
storing the result of the get_from_scope into the dst. Then
we'd emit the TDZ check. The TDZ check can throw, and it could
lead to us returning TDZ from the eval itself. Instead, we need
to use a temporary to emit a TDZ check on. Only after the TDZ check
passes can we move the temporary into the result.
- bytecompiler/NodesCodegen.cpp: (JSC::ShortCircuitReadModifyResolveNode::emitBytecode):
git-svn-id: https://svn.webkit.org/repository/webkit/trunk@278819 268f45cc-cd09-0410-ab3c-d52691b4dbfc
- 2:49 PM Changeset in webkit [278903] by
-
- 3 edits in branches/safari-611-branch/Source/WebCore
Cherry-pick r278755. rdar://problem/79355266
CARingBuffer::frameOffset() makes incorrect assumptions about the frame count
https://bugs.webkit.org/show_bug.cgi?id=226253
<rdar://problem/78463453>
Reviewed by Eric Carlson.
CARingBuffer::frameOffset() was trying to avoid doing a
frameNumber % m_frameCount
modulo operation by doing aframeNumber & (m_frameCount - 1). However, this bitwise
operation is only equivalent if m_frameCount is a power of 2. It isn't enforced
anywhere that the frameCount is a power of 2. As a matter of fact, we frequently use
2*sampleRate which is often 2*44100=88200, which is NOT a power of 2.
When adding logging, I saw frameOffset(512) returning 0 for a frameCount of 88200, which
made no sense. It was causing offset0 and offset1 in CARingBuffer::fetchInternal() to
be both 0 (even though startRead was 0 and endRead was 512) and it was leading the
function to make bad computations.
To address the issue, I updated CARingBuffer::frameOffset() to use a simple modulo
operation. It is safer as it makes no assumption on the frame count and it is more
readable. If we're worried about the performance, we could alternatively round up the
frameCount to the next power of 2 and keep using the bitwise operation, but I am
personally do not think it is worth it.
- platform/audio/cocoa/CARingBuffer.cpp: (WebCore::CARingBuffer::initializeAfterAllocation):
- platform/audio/cocoa/CARingBuffer.h: (WebCore::CARingBuffer::frameOffset):
git-svn-id: https://svn.webkit.org/repository/webkit/trunk@278755 268f45cc-cd09-0410-ab3c-d52691b4dbfc
- 2:49 PM Changeset in webkit [278902] by
-
- 4 edits1 add in branches/safari-611-branch
Cherry-pick r278578. rdar://problem/79355258
Short circuit read modify write nodes emit byte code that uses the wrong locals
https://bugs.webkit.org/show_bug.cgi?id=226576
<rdar://problem/78810362>
Reviewed by Yusuke Suzuki.
JSTests:
- stress/short-circuit-read-modify-should-use-the-write-virtual-registers.js: Added. (eval):
Source/JavaScriptCore:
It's never a good idea to use the wrong local :-)
This patch also adds support for dumping predecessors of basic blocks
in the bytecode dump.
- bytecode/BytecodeDumper.cpp: (JSC::CodeBlockBytecodeDumper<Block>::dumpGraph):
- bytecompiler/NodesCodegen.cpp: (JSC::ShortCircuitReadModifyResolveNode::emitBytecode): (JSC::ShortCircuitReadModifyDotNode::emitBytecode): (JSC::ShortCircuitReadModifyBracketNode::emitBytecode):
git-svn-id: https://svn.webkit.org/repository/webkit/trunk@278578 268f45cc-cd09-0410-ab3c-d52691b4dbfc
- 2:49 PM Changeset in webkit [278901] by
-
- 12 edits in branches/safari-611-branch
Cherry-pick r278410. rdar://problem/79355285
Cherry-pick webrtc fix to correctly handle audio track state in case of renegotiation
https://bugs.webkit.org/show_bug.cgi?id=226577
Reviewed by Eric Carlson.
LayoutTests/imported/w3c:
- web-platform-tests/webrtc/receiver-track-live.https-expected.txt:
Source/ThirdParty/libwebrtc:
Cherry-pick https://webrtc.googlesource.com/src/+/c335b0e63bff56ca0fbfa617dee6a644c85df164%5E%21/.
We need to do small changes to peer_connection.cc given the upstream fix is based on a newer version
which has some code moved from peer_connection.cc to rtp_transmission_manager.cc.
- Source/webrtc/pc/audio_rtp_receiver.cc:
- Source/webrtc/pc/audio_rtp_receiver.h:
- Source/webrtc/pc/peer_connection.cc:
- Source/webrtc/pc/peer_connection_rtp_unittest.cc:
- Source/webrtc/pc/remote_audio_source.cc:
- Source/webrtc/pc/remote_audio_source.h:
- Source/webrtc/pc/rtp_sender_receiver_unittest.cc:
LayoutTests:
Update test now that we have the correct behavior.
- webrtc/receiver-track-should-stay-live-even-if-receiver-is-inactive.html:
git-svn-id: https://svn.webkit.org/repository/webkit/trunk@278410 268f45cc-cd09-0410-ab3c-d52691b4dbfc
- 2:49 PM Changeset in webkit [278900] by
-
- 6 edits in branches/safari-611-branch
Cherry-pick r278318. rdar://problem/79355222
REGRESSION (iOS 14.5): Can't go back and render previous page properly after "location.href"
https://bugs.webkit.org/show_bug.cgi?id=226323
<rdar://problem/78623536>
Reviewed by Alex Christensen.
Source/WebKit:
A while back, we did an optimization to allow several WebPage objects associated with the
same WebPageProxy to live in the same WebProcess. This allowed us to reuse a process from
a SuspendedPageProxy for a forward navigation, without destroying the SuspendedPageProxy.
However, this added quite a bit of complexity and this broke some same-process back/forward
navigations like in this bug. In particular, it is really hard to get do our history
management right (with the current model) if there is more than more WebPage in a process
for the same WebPageProxy.
To address issues, we go back to the older model with one WebPage per WebProcess for a
given WebPageProxy. To achieve this, we make sure to destroy of SuspendedPageProxy objects
for the current page and destination process before we process-swap (like we used to do).
- UIProcess/WebBackForwardCache.cpp: (WebKit::WebBackForwardCache::removeEntriesForPageAndProcess):
- UIProcess/WebBackForwardCache.h:
- UIProcess/WebPageProxy.cpp: (WebKit::WebPageProxy::receivedNavigationPolicyDecision):
Tools:
New API test written by Alex Christensen to cover this case.
- TestWebKitAPI/Tests/WebKitCocoa/ProcessSwapOnNavigation.mm:
git-svn-id: https://svn.webkit.org/repository/webkit/trunk@278318 268f45cc-cd09-0410-ab3c-d52691b4dbfc
- 2:49 PM Changeset in webkit [278899] by
-
- 2 edits in branches/safari-611-branch/Source/ThirdParty/libwebrtc
Cherry-pick r278272. rdar://problem/79355341
Use tighter bitrate allocation rules for WebRTC H264 software encoder
https://bugs.webkit.org/show_bug.cgi?id=226319
<rdar://73150695>
Reviewed by Eric Carlson.
Software H264 encoder is sometimes overshooting target bitrate in which case WebRTC backend will start dropping frames.
The encoder might then think it is on target and will not try to increase compression.
This makes it possible to be locked in a very low frame rate but high quality image situation.
It is often better to preserve frame rate and lower quality, the application could always lower frame rate if desired.
To do so, we detect whether the encoder is using software code path or not.
If so, we compute the actual frame rate and compare it with the expected frame rate.
If the actual frame rate is twice smaller or even below, we enter in a low frame rate mode.
Otherwise, we are in a regular frame rate mode where we apply the normal bitrate rules.
In the low frame rate mode, we divide the target bitrate by 3 so as to quickly recover frame rate.
This works well in situations where motion increases from time to time.
It is still not perfect for instance in case the video is muted and gets unmuted or when the scene is completely still and suddenly large motion happens.
In those cases, frame rate is recovered after a minute or so according my testing.
- Source/webrtc/sdk/objc/components/video_codec/RTCVideoEncoderH264.mm: (-[RTCVideoEncoderH264 initWithCodecInfo:]): (-[RTCVideoEncoderH264 encode:codecSpecificInfo:frameTypes:]): (-[RTCVideoEncoderH264 resetCompressionSessionWithPixelFormat:]): (-[RTCVideoEncoderH264 setEncoderBitrateBps:frameRate:]): (-[RTCVideoEncoderH264 updateBitRateAccordingActualFrameRate]):
git-svn-id: https://svn.webkit.org/repository/webkit/trunk@278272 268f45cc-cd09-0410-ab3c-d52691b4dbfc
- 2:02 PM Changeset in webkit [278898] by
-
- 3 edits in trunk/Source/ThirdParty/ANGLE
Tensorflow.js Broken in Safari 15
https://bugs.webkit.org/show_bug.cgi?id=226953
Tensorflow shaders use a 'NAN' uniform. This is a reserved keyword
in metal. Add NAN to the reserved list.
'inf' constants need to be replaced with INFINITY, much like nan constants need to be replaced with NAN
Reviewed by Kenneth Russell.
- src/compiler/translator/TranslatorMetalDirect.cpp:
(sh::GetMslKeywords):
- 1:57 PM Changeset in webkit [278897] by
-
- 2 edits in trunk/LayoutTests
[WinCairo] Unreviewed test gardening
- platform/wincairo/TestExpectations:
- 1:55 PM Changeset in webkit [278896] by
-
- 5 edits in branches/safari-611-branch/Source/WebKit
Revert r277536. rdar://problem/78875336
This reverts r278882.
- 1:53 PM Changeset in webkit [278895] by
-
- 14 edits in trunk/Source/WebKit
Add basic detection of unresponsive Network / GPU Processes
https://bugs.webkit.org/show_bug.cgi?id=226994
Reviewed by Geoffrey Garen.
If a WebProcess A attempts to connect to the GPUProcess / NetworkProcess B and process B
fails to respond within 3 seconds, we consider process B as unresponsive and terminate
it. As a result, it will re-attempt to launch process B and connect to it again.
This helps in the following scenario:
- User is in a tab and something looks broken due to the GPUProcess or NetworkProcess becoming unresponsive
- The user tries to reload the page in a new tab
Before this patch, the tab would be similarly broken / hung because the GPUProcess or
NetworkProcess would be unresponsive. After this patch, we would detect the process is
hung and kill it. As a result, the page would load correctly in a new tab (with a delay).
In a follow-up, I think we should consider doing the same thing when the user requests
a reload, so that we don't require opening a new tab to recover. However, this keeps
the patch small and is a decent first step.
- Shared/ProcessTerminationReason.h:
- UIProcess/AuxiliaryProcessProxy.cpp:
(WebKit::AuxiliaryProcessProxy::AuxiliaryProcessProxy):
(WebKit::AuxiliaryProcessProxy::didFinishLaunching):
(WebKit::AuxiliaryProcessProxy::shutDownProcess):
(WebKit::AuxiliaryProcessProxy::platformIsBeingDebugged const):
(WebKit::AuxiliaryProcessProxy::stopResponsivenessTimer):
(WebKit::AuxiliaryProcessProxy::startResponsivenessTimer):
(WebKit::AuxiliaryProcessProxy::mayBecomeUnresponsive):
(WebKit::AuxiliaryProcessProxy::didBecomeUnresponsive):
- UIProcess/AuxiliaryProcessProxy.h:
(WebKit::AuxiliaryProcessProxy::responsivenessTimer):
(WebKit::AuxiliaryProcessProxy::responsivenessTimer const):
- UIProcess/Cocoa/WebProcessProxyCocoa.mm:
(WebKit::WebProcessProxy::platformIsBeingDebugged const): Deleted.
- UIProcess/GPU/GPUProcessProxy.cpp:
(WebKit::GPUProcessProxy::getGPUProcessConnection):
(WebKit::GPUProcessProxy::gpuProcessExited):
(WebKit::GPUProcessProxy::didBecomeUnresponsive):
- UIProcess/GPU/GPUProcessProxy.h:
- UIProcess/Network/NetworkProcessProxy.cpp:
(WebKit::NetworkProcessProxy::didBecomeUnresponsive):
(WebKit::NetworkProcessProxy::getNetworkProcessConnection):
- UIProcess/Network/NetworkProcessProxy.h:
- UIProcess/ResponsivenessTimer.cpp:
(WebKit::ResponsivenessTimer::timerFired):
(WebKit::ResponsivenessTimer::mayBecomeUnresponsive const):
- UIProcess/ResponsivenessTimer.h:
- UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::gpuProcessExited):
- UIProcess/WebProcessProxy.cpp:
(WebKit::WebProcessProxy::WebProcessProxy):
(WebKit::WebProcessProxy::shutDown):
(WebKit::WebProcessProxy::didFinishLaunching):
(WebKit::WebProcessProxy::isResponsive const):
(WebKit::WebProcessProxy::processTerminated):
(WebKit::WebProcessProxy::platformIsBeingDebugged const): Deleted.
(WebKit::WebProcessProxy::mayBecomeUnresponsive): Deleted.
(WebKit::WebProcessProxy::stopResponsivenessTimer): Deleted.
(WebKit::WebProcessProxy::startResponsivenessTimer): Deleted.
- UIProcess/WebProcessProxy.h:
(WebKit::WebProcessProxy::responsivenessTimer): Deleted.
- 1:19 PM Changeset in webkit [278894] by
-
- 4 edits in trunk
Allow legacy SecurityOrigin behavior for x-apple-ql-id2 scheme
https://bugs.webkit.org/show_bug.cgi?id=226993
Source/WebCore:
<rdar://76474042>
Reviewed by Tim Horton.
Why have one x-apple-ql-id scheme when you can have 2?
- page/SecurityOrigin.cpp:
(WebCore::shouldTreatAsUniqueOrigin):
Also move the linked-on-or-before check to after the hasSpecialScheme check which will be true for most URLs (http, https, file, etc.)
Tools:
Reviewed by Tim Horton.
- TestWebKitAPI/Tests/WebCore/SecurityOrigin.cpp:
(TestWebKitAPI::TEST_F):
- 1:12 PM Changeset in webkit [278893] by
-
- 2 edits in trunk/Source/ThirdParty/ANGLE
Fix typo in r276318
https://bugs.webkit.org/show_bug.cgi?id=227041
<rdar://76284889>
Reviewed by Chris Dumez.
@ uses the string @(DYLIB_INSTALL_NAME_BASE)
$ uses the value of the environment variable.
This turns out to be an important difference.
- Configurations/ANGLE-dynamic.xcconfig:
- 1:11 PM Changeset in webkit [278892] by
-
- 2 edits in trunk/Source/WebCore
[GStreamer] Another crash under gst_element_add_pad
https://bugs.webkit.org/show_bug.cgi?id=225765
Patch by Philippe Normand <pnormand@igalia.com> on 2021-06-15
Reviewed by Adrian Perez de Castro.
Decodebin3 in GStreamer <= 1.16 does not respect user-supplied select-stream events. So we
need to relax the release assert for these versions. This bug was fixed in:
https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/commit/b41b87522f59355bb21c001e9e2df96dc6956928
- platform/graphics/gstreamer/ImageDecoderGStreamer.cpp:
(WebCore::ImageDecoderGStreamer::InnerDecoder::connectDecoderPad):
- 12:15 PM Changeset in webkit [278891] by
-
- 6 edits4 deletes in trunk/Source
Delete some recently-obsoleted files
https://bugs.webkit.org/show_bug.cgi?id=226705
Reviewed by Chris Dumez.
Source/WebCore:
- Sources.txt: Removed CSSCustomIdentValue.cpp.
- WebCore.xcodeproj/project.pbxproj: Removed CSSCustomIdentValue.cpp/h.
- css/CSSCustomIdentValue.cpp: Removed.
- css/CSSCustomIdentValue.h: Removed.
- css/calc/CSSCalcExpressionNodeParser.cpp: Added TextStream.h include.
- dom/SuccessOr.h: Removed.
Source/WTF:
- wtf/CMakeLists.txt: Removed Optional.h.
- wtf/Optional.h: Removed.
- 12:05 PM Changeset in webkit [278890] by
-
- 7 edits in trunk/Tools
[git-webkit] Handle auth failures
https://bugs.webkit.org/show_bug.cgi?id=226999
<rdar://problem/79313850>
Reviewed by Dewei Zhu.
- Scripts/libraries/webkitcorepy/setup.py: Bump version.
- Scripts/libraries/webkitcorepy/webkitcorepy/init.py: Ditto.
- Scripts/libraries/webkitcorepy/webkitcorepy/credentials.py:
(credentials): We should attempt to retrieve credentials, even if they
are not requested because retrieving credentials does not prompt the user.
- Scripts/libraries/webkitscmpy/setup.py: Bump version.
- Scripts/libraries/webkitscmpy/webkitscmpy/init.py: Ditto.
- Scripts/libraries/webkitscmpy/webkitscmpy/remote/git_hub.py:
(GitHub.request): In the event we get a 400 error from the GitHub API,
attempt forcing authentication before reporting an error.
- 12:04 PM Changeset in webkit [278889] by
-
- 2 edits in trunk/Source/WebKit
[Monterey] Support building WebKit (Follow-up)
https://bugs.webkit.org/show_bug.cgi?id=226846
<rdar://problem/79095148>
Unreviewed follow-up fix.
Covered by exisiting tests.
- WebKit.xcodeproj/project.pbxproj: Reset objectVersion.
- 11:39 AM Changeset in webkit [278888] by
-
- 6 edits in trunk/Source/JavaScriptCore
Shouldn't drain the micro task queue when calling out to ObjC
https://bugs.webkit.org/show_bug.cgi?id=161942
Unreviewed, relanding r278734.
- API/tests/testapi.cpp:
(TestAPI::promiseDrainDoesNotEatExceptions):
(testCAPIViaCpp):
- API/tests/testapi.mm:
(testMicrotaskWithFunction):
(testObjectiveCAPI):
- runtime/JSLock.cpp:
(JSC::JSLock::willReleaseLock):
- runtime/ObjectPrototype.cpp:
(JSC::isPokerBros):
- runtime/VM.cpp:
(JSC::VM::didExhaustMicrotaskQueue):
- 11:25 AM Changeset in webkit [278887] by
-
- 2 edits in trunk/Source/WebCore
[GPU Process] Add a bounds check before reading data length for out-of-line display list items
https://bugs.webkit.org/show_bug.cgi?id=227029
rdar://79343645
Reviewed by Chris Dumez.
Add some additional hardening when decoding out-of-line display list items.
- platform/graphics/displaylists/DisplayListIterator.cpp:
(WebCore::DisplayList::DisplayList::Iterator::updateCurrentItem):
Currently, it's possible to perform an out-of-bounds read in the shared display list item buffer, since we only
perform a bounds check after grabbing the data length (8 bytes) from the buffer after reading and validating an
out-of-line display list item type.
Mitigate this by validating that there is enough buffer capacity for both the padded item type value and encoded
data length, before we attempt to read the encoded data length.
- 11:21 AM Changeset in webkit [278886] by
-
- 2 edits in trunk/Source/ThirdParty/libwebrtc
REGRESSION: [ BigSur iOS wk1 Debug ] imported/w3c/web-platform-tests/html/dom/usvstring-reflection.https.html is flaky crashing
https://bugs.webkit.org/show_bug.cgi?id=226928
<rdar://problem/79205526>
Reviewed by Eric Carlson.
- Source/webrtc/rtc_base/task_queue_stdlib.cc:
thread_ is created in TaskQueueStdlib constructor before other members pending_queue_ or delayed_queue_.
It might happen that thread_ will start executing its callback which will read delayed_queue_ before it is initialized by TaskQueueStdlib constructor.
Update to create thread_ before all other members.
- 11:18 AM Changeset in webkit [278885] by
-
- 3 edits in trunk/Source/WebCore
GraphicsContext::apply{Fill,Stroke}Pattern needs to do nothing if there is no fill/stroke pattern set
https://bugs.webkit.org/show_bug.cgi?id=227015
<rdar://problem/79301173>
Reviewed by Simon Fraser.
Now that we have a GPU Process, we can't trust incoming IPC. So, this patch protects
against trying to apply the curretn fill/stroke pattern when there isn't one set.
- platform/graphics/cg/GraphicsContextCG.cpp:
(WebCore::GraphicsContextCG::applyStrokePattern):
(WebCore::GraphicsContextCG::applyFillPattern):
- platform/graphics/win/GraphicsContextDirect2D.cpp:
(WebCore::GraphicsContextDirect2D::applyStrokePattern):
(WebCore::GraphicsContextDirect2D::applyFillPattern):
- 11:15 AM Changeset in webkit [278884] by
-
- 6 edits in trunk/Source/WebCore
GraphicsContext restores need to do nothing if the state stack is empty
https://bugs.webkit.org/show_bug.cgi?id=227009
<rdar://problem/79301368>
Reviewed by Simon Fraser.
Now that we have a GPU Process, we can't trust incoming IPC. So, this patch protects
against unbalanced save/restore commands in GraphicsContexts.
- platform/graphics/cairo/GraphicsContextCairo.cpp:
(WebCore::GraphicsContextCairo::restore):
- platform/graphics/cairo/PlatformContextCairo.cpp:
(WebCore::PlatformContextCairo::restore):
- platform/graphics/cg/GraphicsContextCG.cpp:
(WebCore::GraphicsContextCG::restore):
- platform/graphics/nicosia/cairo/NicosiaCairoOperationRecorder.cpp:
(Nicosia::CairoOperationRecorder::restore):
- platform/graphics/win/GraphicsContextDirect2D.cpp:
(WebCore::GraphicsContextDirect2D::restore):
- 11:12 AM Changeset in webkit [278883] by
-
- 25 edits2 adds in trunk
Scrollbars are not clipped to layer bounds in RenderLayerBacking
https://bugs.webkit.org/show_bug.cgi?id=226823
Reviewed by Simon Fraser.
Source/WebCore:
Mask contents to bounds of overflow controls containers in RenderLayerBacking.
Test: compositing/scrolling/async-overflow-scrolling/overflow-controls-container-clips-scrollbars.html
- rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::updateGeometry):
(WebCore::RenderLayerBacking::adjustOverflowControlsPositionRelativeToAncestor):
LayoutTests:
Adjust test expectations due to clipping on overflow controls container.
- compositing/clipping/border-radius-async-overflow-clipping-layer-expected.txt:
- compositing/layer-creation/clipping-scope/nested-scroller-overlap-expected.txt:
- compositing/rtl/rtl-scrolling-with-transformed-descendants-expected.txt:
- compositing/scrolling/async-overflow-scrolling/clipped-layer-in-overflow-clipped-by-scroll-expected.txt:
- compositing/scrolling/async-overflow-scrolling/clipped-layer-in-overflow-expected.txt:
- compositing/scrolling/async-overflow-scrolling/clipped-layer-in-overflow-nested-expected.txt:
- compositing/scrolling/async-overflow-scrolling/layer-for-negative-z-in-scroller-expected.txt:
- compositing/scrolling/async-overflow-scrolling/layer-in-overflow-clip-to-hidden-expected.txt:
- compositing/scrolling/async-overflow-scrolling/layer-in-overflow-clip-to-visible-expected.txt:
- compositing/scrolling/async-overflow-scrolling/layer-in-overflow-expected.txt:
- compositing/scrolling/async-overflow-scrolling/layer-in-overflow-gain-clipping-layer-expected.txt:
- compositing/scrolling/async-overflow-scrolling/layer-in-overflow-in-clipped-expected.txt:
- compositing/scrolling/async-overflow-scrolling/layer-in-overflow-lose-clipping-layer-expected.txt:
- compositing/scrolling/async-overflow-scrolling/nested-scrollers-backing-attachment-expected.txt:
- compositing/scrolling/async-overflow-scrolling/overflow-controls-container-clips-scrollbars-expected.html: Added.
- compositing/scrolling/async-overflow-scrolling/overflow-controls-container-clips-scrollbars.html: Added.
- compositing/scrolling/async-overflow-scrolling/transform-change-scrollbar-position-expected.txt:
- compositing/scrolling/async-overflow-scrolling/visibility-hidden-scrollers-expected.txt:
- compositing/shared-backing/overflow-scroll/previous-sibling-prevents-inclusiveness-expected.txt:
- compositing/tiling/perspective-on-scroller-tile-coverage-expected.txt:
- platform/mac-wk1/compositing/clipping/border-radius-async-overflow-clipping-layer-expected.txt:
- tiled-drawing/scrolling/overflow/overflow-scrolled-down-tile-coverage-expected.txt:
- tiled-drawing/scrolling/overflow/overflow-scrolled-up-tile-coverage-expected.txt:
- tiled-drawing/scrolling/overflow/overflow-tile-coverage-expected.txt:
- 11:09 AM Changeset in webkit [278882] by
-
- 5 edits in branches/safari-611-branch/Source/WebKit
Cherry-pick r277536. rdar://problem/78875336
This reverts r278529.
- 11:01 AM Changeset in webkit [278881] by
-
- 3 edits in trunk/Source/WebKit
[iOS] Permit access to 'platform-name'
https://bugs.webkit.org/show_bug.cgi?id=227032
<rdar://problem/79334360>
Reviewed by Per Arne Vollan.
Telemetry shows that some media code needs access to 'platform-name'. We allow this in the
WebContent process, but didn't include it in the GPU Process.
Make the WebContent and GPU Process consistent in this access. Also adopt the standard rule
for this used in the system sandbox, which restricts the access to the IOKit class family that
has this property.
- Resources/SandboxProfiles/ios/com.apple.WebKit.GPU.sb:
- Resources/SandboxProfiles/ios/com.apple.WebKit.WebContent.sb.in:
- 10:20 AM Changeset in webkit [278880] by
-
- 25 edits in trunk/Source/WebKit
[GTK] Support transient zoom
https://bugs.webkit.org/show_bug.cgi?id=197002
Patch by Alexander Mikhaylenko <Alexander Mikhaylenko> on 2021-06-15
Reviewed by Michael Catanzaro.
Speed up pinch zoom for AC mode. Reuse the same infrastructure as the
macOS implementation. Move the shared code to ViewGestureController.cpp
where possible.
Slightly adjust the shared code to support changing the gesture origin
during the gesture since we support it on both touchscreen and touchpad.
There's not much we can do for non-AC mode so just keep the current
behavior there, wrapping real-time zoom into transient zoom API so that
it's confined to the web process.
- UIProcess/API/gtk/WebKitWebViewBase.cpp:
(webkitWebViewBaseZoomChanged):
(webkitWebViewBaseZoomEnd):
(webkitWebViewBaseConstructed):
(webkitWebViewBaseZoomBegin): Deleted.
- UIProcess/CoordinatedGraphics/DrawingAreaProxyCoordinatedGraphics.cpp:
(WebKit::DrawingAreaProxyCoordinatedGraphics::adjustTransientZoom):
(WebKit::DrawingAreaProxyCoordinatedGraphics::commitTransientZoom):
- UIProcess/CoordinatedGraphics/DrawingAreaProxyCoordinatedGraphics.h:
- UIProcess/ViewGestureController.cpp:
(WebKit::ViewGestureController::scaledMagnificationOrigin):
Support changing origin during the gesture, cache the page scale factor
to accommodate non-AC mode doing real-time zooming instead of transient
zooming.
(WebKit::ViewGestureController::didCollectGeometryForMagnificationGesture):
(WebKit::ViewGestureController::prepareMagnificationGesture):
(WebKit::ViewGestureController::applyMagnification):
(WebKit::ViewGestureController::endMagnificationGesture):
(WebKit::ViewGestureController::magnification const):
Moved from ViewGestureControllerMac.mm.
- UIProcess/ViewGestureController.h:
- UIProcess/ViewGestureController.messages.in:
- UIProcess/WebPageProxy.h:
- UIProcess/gtk/ViewGestureControllerGtk.cpp:
(WebKit::ViewGestureController::setMagnification):
(WebKit::ViewGestureController::endMagnification):
- UIProcess/gtk/WebPageProxyGtk.cpp:
(WebKit::WebPageProxy::getCenterForZoomGesture): Deleted.
- UIProcess/mac/ViewGestureControllerMac.mm:
(WebKit::ViewGestureController::resistanceForDelta):
Make it a static private method instead of a function so that we can
keep using minMagnification and maxMagnification in it.
(WebKit::ViewGestureController::handleMagnificationGestureEvent):
(WebKit::ViewGestureController::didCollectGeometryForSmartMagnificationGesture):
Set the initial scale and origin before calling scaledMagnificationOrigin().
(WebKit::resistanceForDelta): Deleted.
(WebKit::ViewGestureController::scaledMagnificationOrigin):
(WebKit::ViewGestureController::didCollectGeometryForMagnificationGesture):
(WebKit::ViewGestureController::endMagnificationGesture):
(WebKit::ViewGestureController::magnification const):
Moved to ViewGestureController.cpp.
- WebProcess/WebPage/CoordinatedGraphics/CompositingCoordinator.cpp:
(WebKit::CompositingCoordinator::flushPendingLayerChanges):
We only need to set the ApplyScrollingTreeLayerPositions flag when not
zooming, so add a parameter to do that.
- WebProcess/WebPage/CoordinatedGraphics/CompositingCoordinator.h:
- WebProcess/WebPage/CoordinatedGraphics/DrawingAreaCoordinatedGraphics.cpp:
(WebKit::DrawingAreaCoordinatedGraphics::adjustTransientZoom):
(WebKit::DrawingAreaCoordinatedGraphics::commitTransientZoom):
- WebProcess/WebPage/CoordinatedGraphics/DrawingAreaCoordinatedGraphics.h:
- WebProcess/WebPage/CoordinatedGraphics/LayerTreeHost.cpp:
(WebKit::LayerTreeHost::layerFlushTimerFired):
(WebKit::LayerTreeHost::forceRepaint):
Don't pass ApplyScrollingTreeLayerPositions when transient zoom is active.
(WebKit::LayerTreeHost::constrainTransientZoomOrigin const):
(WebKit::LayerTreeHost::layerForTransientZoom const):
(WebKit::LayerTreeHost::applyTransientZoomToLayers):
(WebKit::LayerTreeHost::adjustTransientZoom):
(WebKit::LayerTreeHost::commitTransientZoom):
- WebProcess/WebPage/CoordinatedGraphics/LayerTreeHost.h:
- WebProcess/WebPage/DrawingArea.h:
(WebKit::DrawingArea::addTransactionCallbackID):
(WebKit::DrawingArea::commitTransientZoom):
- WebProcess/WebPage/DrawingArea.messages.in:
- WebProcess/WebPage/ViewGestureGeometryCollector.cpp:
(WebKit::ViewGestureGeometryCollector::collectGeometryForMagnificationGesture):
- WebProcess/WebPage/ViewGestureGeometryCollector.h:
- WebProcess/WebPage/ViewGestureGeometryCollector.messages.in:
Build CollectGeometryForMagnificationGesture for GTK as well.
- WebProcess/WebPage/WebPage.h:
- WebProcess/WebPage/WebPage.messages.in:
- WebProcess/WebPage/gtk/WebPageGtk.cpp:
(WebKit::WebPage::getCenterForZoomGesture): Deleted.
- 9:59 AM Changeset in webkit [278879] by
-
- 18 edits in trunk
Check for "xn--" in any subdomain when parsing URL hosts
https://bugs.webkit.org/show_bug.cgi?id=226912
Reviewed by Darin Adler.
LayoutTests/imported/w3c:
- web-platform-tests/url/a-element-expected.txt:
- web-platform-tests/url/a-element-xhtml-expected.txt:
- web-platform-tests/url/failure-expected.txt:
- web-platform-tests/url/resources/urltestdata.json:
- web-platform-tests/url/toascii.window-expected.txt:
- web-platform-tests/url/url-constructor-expected.txt:
Source/WTF:
We have a fast path that doesn't call uidna_nameToASCII if the host is already ASCII.
We need to check if the host is properly-punycode-encoded if it starts with "xn--"
but we also need to check if any subdomain starts with "xn--" (not just the first one).
In order to not regress tests, I needed to also take the fix I did in r256629 and apply it to all use of uidna_nameToASCII.
- wtf/URL.cpp:
(WTF::appendEncodedHostname):
- wtf/URLHelpers.cpp:
(WTF::URLHelpers::mapHostName):
- wtf/URLParser.cpp:
(WTF::URLParser::domainToASCII):
(WTF::URLParser::subdomainStartsWithXNDashDash):
(WTF::URLParser::parseHostAndPort):
(WTF::URLParser::startsWithXNDashDash): Deleted.
- wtf/URLParser.h:
Tools:
- TestWebKitAPI/Tests/WTF/URLParser.cpp:
(TestWebKitAPI::TEST_F):
These tests used to hit UIDNA_ERROR_LABEL_TOO_LONG which is allowed now.
- TestWebKitAPI/Tests/WTF/cocoa/URLExtras.mm:
(TestWebKitAPI::TEST):
This test, from r262171, needs to verify that non-ASCII characters are not truncated to ASCII values when converting to NSURL.
It used to use an invalid URL that had a host that ended in U+FE63 (SMALL HYPHEN-MINUS) which would fail because of UIDNA_ERROR_TRAILING_HYPHEN.
Now that trailing hyphens are allowed, we end in U+0661 and U+06F1 which fail because of UIDNA_ERROR_BIDI which makes this test still verify
the non-truncated values of an invalid host converted to an NSURL.
LayoutTests:
- fast/dom/DOMURL/parsing-expected.txt:
- fast/dom/DOMURL/parsing.html:
Update the test I added in r236527 to reflect this relaxation.
This matches the behavior of Chrome Canary.
- 9:47 AM Changeset in webkit [278878] by
-
- 6 edits in trunk/Source
-Warray-bounds warning in Packed.h
https://bugs.webkit.org/show_bug.cgi?id=226557
<rdar://problem/79103658>
Patch by Michael Catanzaro <Michael Catanzaro> on 2021-06-15
Reviewed by Darin Adler.
Source/JavaScriptCore:
- b3/air/AirAllocateRegistersByGraphColoring.cpp:
- jit/JITCall.cpp:
(JSC::JIT::compileOpCall):
Source/WTF:
- wtf/Bitmap.h:
(WTF::WordType>::clear): Use the newly-introduced IGNORE_ARRAY_BOUNDS_WARNINGS macros
instead of vanilla IGNORE_WARNINGS.
- wtf/Compiler.h: Add new IGNORE_ARRAY_BOUNDS_WARNINGS_[BEGIN,END] macros, since this
warning is now suppressed in several different places.
- wtf/Packed.h: Suppress the warning. Also, add a static_assert for safety.
- 9:46 AM Changeset in webkit [278877] by
-
- 15 edits in trunk/Source
[Cocoa] Harden WebAuthn process by restricting to browser-entitled processes
https://bugs.webkit.org/show_bug.cgi?id=226772
<rdar://problem/74721877>
Reviewed by Darin Adler.
Source/WebKit:
We should ensure that any process attempting to launch the WebAuthn XPC service is entitled as a full web browser. We
should also ensure that the process requesting WebAuthn operations over WebKit CoreIPC is the WebContent process.
Tested by existing WebAuthn tests.
- Platform/Logging.h: Add WebAuthn category.
- Shared/Cocoa/DefaultWebBrowserChecks.mm:
(WebKit::isInWebKitChildProcess): Recognize 'com.apple.WebKit.WebAuthn' as a WebKit sub-process.
- UIProcess/Cocoa/WebProcessProxyCocoa.mm:
(WebKit::WebProcessProxy::hasCorrectPACEntitlement): Added.
- UIProcess/WebProcessProxy.cpp:
(WebKit::WebProcessProxy::getWebAuthnProcessConnection): Add a new MESSAGE_CHECK that the process requesting a WebAuthn
process connection is a WebContent process.
- UIProcess/WebProcessProxy.h:
- WebAuthnProcess/WebAuthnProcess.cpp:
(WebKit::WebAuthnProcess::platformInitializeWebAuthnProcess): Add some logging.
- WebAuthnProcess/WebAuthnProcess.h:
- WebAuthnProcess/ios/WebAuthnProcessIOS.mm:
(WebKit::WebAuthnProcess::platformInitializeWebAuthnProcess): Ensure that the application invoking the WebAuthn XPC service
is entitled as a default web browser.
- WebAuthnProcess/mac/WebAuthnProcessMac.mm:
(WebKit::WebAuthnProcess::platformInitializeWebAuthnProcess): Stub.
- WebProcess/WebProcess.cpp:
(WebKit::getWebAuthnProcessConnection):
(WebKit::WebProcess::ensureWebAuthnProcessConnection):
Source/WTF:
Add a new entitlement utility that returns the String value of a given entitlement for
the process with the provided audit token.
- wtf/cocoa/Entitlements.h:
- wtf/cocoa/Entitlements.mm:
(WTF::hasEntitlementValue):
- wtf/PlatformHave.h: Add item for PAC shared region feature.
- 9:41 AM Changeset in webkit [278876] by
-
- 8 edits in branches/safari-612.1.18.1-branch/Source
Versioning.
WebKit-7612.1.18.1.2
- 9:09 AM Changeset in webkit [278875] by
-
- 14 edits in trunk/Source/JavaScriptCore
Move setting of scratch buffer active lengths to the runtime functions.
https://bugs.webkit.org/show_bug.cgi?id=227013
rdar://79325068
Reviewed by Keith Miller.
We previously emit JIT'ed code to set and unset the ScratchBuffer active length
around calls into C++ runtime functions. This was needed because the runtime
functions may allow GC to run, and GC needs to be able to scan the values stored
in the ScratchBuffer.
In this patch, we change it so that the runtime functions that need it will
declare an ActiveScratchBufferScope RAII object that will set the ScratchBuffer
active length, and unset it on exit. This allows us to:
- Emit less JIT code. The runtime function can take care of it.
- Elide setting the ScratchBuffer active length if not needed. The runtime functions know whether they can GC or not. They only need to set the active length if they can GC.
Note that scanning of the active ScratchBuffer is done synchronously on the
mutator thread via Heap::gatherScratchBufferRoots(), which is called as part of
the GC conservative root scan. This means there is no urgency / sequencing that
requires that the active length be set before calling into the runtime function.
Setting it in the runtime function itself is fine as long as it is done before
the function executes any operations that can GC.
This patch also made the following changes:
- Introduce ActiveScratchBufferScope RAII object used to set/unset the ScratchBuffer length in the runtime functions. ActiveScratchBufferScope takes the active length in units of number of stack slots / Registers / JSValues instead of bytes.
- Deleted ScratchRegisterAllocator::preserveUsedRegistersToScratchBufferForCall() and ScratchRegisterAllocator::restoreUsedRegistersFromScratchBufferForCall(). These functions are unused.
The reasoning behind what values to pass to ActiveScratchBufferScope, is any:
- AssemblyHelpers::debugCall() in AssemblyHelpers.cpp: The ScratchBuffer is only used for operationDebugPrintSpeculationFailure(), which now declares an ActiveScratchBufferScope.
The active length is GPRInfo::numberOfRegisters + FPRInfo::numberOfRegisters.
See scratchSize in AssemblyHelpers::debugCall().
- genericGenerationThunkGenerator() in FTLThunks.cpp: The scratch buffer size for determining the active length is requiredScratchMemorySizeInBytes().
However, genericGenerationThunkGenerator() generates code to call either
operationCompileFTLOSRExit() or operationCompileFTLLazySlowPath(). Both of
these functions will DeferGCForAWhile. Hence, GC cannot run, and we don't need
to set the active length here.
- compileArrayPush() in FTLLowerDFGToB3.cpp:
Cases Array::Int32, Array::Contiguous, or Array::Double calls
operationArrayPushMultiple() or operationArrayPushDoubleMultiple().
For operationArrayPushMultiple(), the active length is elementCount. See
computation of scratchSize.
For operationArrayPushDoubleMultiple(), we don't need to set the active length
because the ScratchBuffer only contains double values. The GC does not need
to scan those.
Case Array::ArrayStorage calls operationArrayPushMultiple().
The active length is elementCount. See computation of scratchSize.
compileNewArray() in FTLLowerDFGToB3.cpp:
Calls operationNewArray(). Active length is m_node->numChildren(), which is
passed to operationNewArray() as the size parameter. See computation of
scratchSize.
compileNewArrayWithSpread() in FTLLowerDFGToB3.cpp:
Calls operationNewArrayWithSpreadSlow(). Active length is m_node->numChildren(),
which is passes to operationNewArrayWithSpreadSlow() as the numItems parameter.
See computation of scratchSize.
- osrExitGenerationThunkGenerator() in DFGThunks.cpp:
Calls operationCompileOSRExit(). Active length is GPRInfo::numberOfRegisters +
FPRInfo::numberOfRegisters. See computation of scratchSize.
- compileNewArray() in DFGSpeculativeJIT.cpp:
Calls operationNewArray(). Active length is node->numChildren(), which is
passed in as the size parameter.
compileNewArrayWithSpread() in DFGSpeculativeJIT.cpp:
Calls operationNewArrayWithSpreadSlow(). Active length is node->numChildren(),
which is passed in as the numItems parameter.
compileArrayPush() in DFGSpeculativeJIT.cpp:
Calls operationArrayPushMultiple(). Active length is elementCount, which is
passed in as the elementCount parameter.
Calls operationArrayPushDoubleMultiple(). Active length is elementCount, but
we don't need to set it because the ScratchBuffer only contains double values.
- dfg/DFGOSRExit.cpp:
(JSC::DFG::JSC_DEFINE_JIT_OPERATION):
- dfg/DFGOperations.cpp:
(JSC::DFG::JSC_DEFINE_JIT_OPERATION):
- dfg/DFGSpeculativeJIT.cpp:
(JSC::DFG::SpeculativeJIT::compileNewArray):
(JSC::DFG::SpeculativeJIT::compileNewArrayWithSpread):
(JSC::DFG::SpeculativeJIT::compileArrayPush):
- dfg/DFGThunks.cpp:
(JSC::DFG::osrExitGenerationThunkGenerator):
- ftl/FTLLowerDFGToB3.cpp:
(JSC::FTL::DFG::LowerDFGToB3::compileArrayPush):
(JSC::FTL::DFG::LowerDFGToB3::compileNewArray):
(JSC::FTL::DFG::LowerDFGToB3::compileNewArrayWithSpread):
- ftl/FTLOSRExitCompiler.cpp:
(JSC::FTL::JSC_DEFINE_JIT_OPERATION):
- ftl/FTLOperations.cpp:
(JSC::FTL::JSC_DEFINE_JIT_OPERATION):
- ftl/FTLThunks.cpp:
(JSC::FTL::genericGenerationThunkGenerator):
- jit/AssemblyHelpers.cpp:
(JSC::AssemblyHelpers::debugCall):
- jit/ScratchRegisterAllocator.cpp:
(JSC::ScratchRegisterAllocator::preserveUsedRegistersToScratchBufferForCall): Deleted.
(JSC::ScratchRegisterAllocator::restoreUsedRegistersFromScratchBufferForCall): Deleted.
- jit/ScratchRegisterAllocator.h:
- runtime/VM.h:
- runtime/VMInlines.h:
(JSC::ActiveScratchBufferScope::ActiveScratchBufferScope):
(JSC::ActiveScratchBufferScope::~ActiveScratchBufferScope):
- 7:47 AM Changeset in webkit [278874] by
-
- 2 edits in trunk/LayoutTests
(r278359) fast/mediastream/mediaElement-gc.html is a flaky failure
https://bugs.webkit.org/show_bug.cgi?id=226991
<rdar://problem/79309901>
Reviewed by Eric Carlson.
- fast/mediastream/mediaElement-gc.html:
Call GC several times if needed to collect the media element.
- 7:47 AM Changeset in webkit [278873] by
-
- 2 edits in trunk/Source/WebKit
Caches::removeCacheEntry should check for m_storage
https://bugs.webkit.org/show_bug.cgi?id=227022
<rdar://79225216>
Reviewed by Antti Koivisto.
removeCacheEntry can be called asynchronoulsy in Cache::open for badly formatted content.
In that case, given this is done asynchronously, m_storage might be nullified.
Add a nullptr check to cope with this.
- NetworkProcess/cache/CacheStorageEngineCaches.cpp:
(WebKit::CacheStorage::Caches::removeCacheEntry):
- 7:14 AM Changeset in webkit [278872] by
-
- 7 edits17 adds in trunk
Source/WebCore/PAL:
[iOS 15] Support building WebKit
https://bugs.webkit.org/show_bug.cgi?id=227002
<rdar://problem/79316598>
Reviewed by Tim Horton.
- pal/spi/ios/QuickLookSPI.h:
Source/WebKit:
[iOS 15] Support building WebKit
https://bugs.webkit.org/show_bug.cgi?id=227002
<rdar://problem/79316598>
Reviewed by Tim Horton.
Covered by existing tests.
- Platform/spi/ios/UIKitSPI.h: Add _insertTextFromCamera.
Tools:
[iOS 15] Support building WebKit
https://bugs.webkit.org/show_bug.cgi?id=227002
<rdar://problem/79316598>
Reviewed by Tim Horton.
- TestWebKitAPI/Tests/WebKitCocoa/EventAttribution.mm: Remove unneeded import.
WebKitLibraries:
[iOS 15] Support building WebKit
https://bugs.webkit.org/show_bug.cgi?id=227002
<rdar://problem/79316598>
Reviewed by Tim Horton.
- WebKitPrivateFrameworkStubs/iOS/15/AppSupport.framework/AppSupport.tbd: Added.
- WebKitPrivateFrameworkStubs/iOS/15/AuthKit.framework/AuthKit.tbd: Added.
- WebKitPrivateFrameworkStubs/iOS/15/BackBoardServices.framework/BackBoardServices.tbd: Added.
- WebKitPrivateFrameworkStubs/iOS/15/CorePrediction.framework/CorePrediction.tbd: Added.
- WebKitPrivateFrameworkStubs/iOS/15/GraphicsServices.framework/GraphicsServices.tbd: Added.
- WebKitPrivateFrameworkStubs/iOS/15/RunningBoardServices.framework/RunningBoardServices.tbd: Added.
- WebKitPrivateFrameworkStubs/iOS/15/SafariSafeBrowsing.framework/SafariSafeBrowsing.tbd: Added.
- WebKitPrivateFrameworkStubs/iOS/15/URLFormatting.framework/URLFormatting.tbd: Added.
- 7:12 AM Changeset in webkit [278871] by
-
- 2 edits in trunk/LayoutTests
[GTK] Unreviewed test gardening: add expected failure
The fast/text/letter-spacing-produces-nan-width.html test crashes on debug.
Patch by Arcady Goldmints-Orlov <Arcady Goldmints-Orlov> on 2021-06-15
- platform/gtk/TestExpectations:
- 6:13 AM Changeset in webkit [278870] by
-
- 4 edits in trunk/Source
[Mac] CoreMedia WrapperClass does not need alignment fixup
https://bugs.webkit.org/show_bug.cgi?id=226978
rdar://78864290
Reviewed by Geoffrey Garen and Dan Bernstein.
Source/WebCore:
- platform/graphics/cocoa/WebCoreDecompressionSession.mm:
(WebCore::WebCoreDecompressionSession::enqueueSample): static_assert CMBufferQueueRef
callback struct alignment and version size.
Source/WebKit:
- Shared/mac/MediaFormatReader/CoreMediaWrapped.h:
(WebKit::CoreMediaWrapped<Wrapped>::vTable): Don't add padding to WrapperClass,
it isn't necessary. static_assert CMBaseClass and WrapperClass alignment and
.version sizes.
- 6:11 AM Changeset in webkit [278869] by
-
- 3 edits2 adds in trunk
[LFC][TFC] Add support for preferred width computation when mixed width types are present
https://bugs.webkit.org/show_bug.cgi?id=227010
Reviewed by Antti Koivisto.
Source/WebCore:
This is in preparation for supporting available space distribution with mixed width types (e.g. percent and fixed).
While this patch only addresses the preferred width computation for the table, certain values already work with the existing distribution code.
Test: fast/layoutformattingcontext/table-space-distribution-mixed-width-type-simple.html
- layout/formattingContexts/table/TableFormattingContext.cpp:
(WebCore::Layout::TableFormattingContext::computedPreferredWidthForColumns):
LayoutTests:
- fast/layoutformattingcontext/table-space-distribution-mixed-width-type-simple-expected.html: Added.
- fast/layoutformattingcontext/table-space-distribution-mixed-width-type-simple.html: Added.
- 4:09 AM Changeset in webkit [278868] by
-
- 4 edits2 adds in trunk
[css-scroll-snap] Scroll snap offsets are interpreted as scroll positions in ScrollableArea
https://bugs.webkit.org/show_bug.cgi?id=226572
Patch by Martin Robinson <mrobinson@igalia.com> on 2021-06-15
Reviewed by Simon Fraser.
Source/WebCore:
Test: css3/scroll-snap/scroll-snap-right-to-left-initial-snapping.html
This is also covered by existing WPT tests, but these tests are composite reference
tests where there are still some failing elements.
- platform/ScrollableArea.cpp:
(WebCore::ScrollableArea::nearestActiveSnapPoint): Use scroll offsets instead of positions.
(WebCore::ScrollableArea::updateScrollSnapState): Ditto.
LayoutTests:
- TestExpectations: Mark test as newly passing.
- css3/scroll-snap/scroll-snap-right-to-left-initial-snapping-expected.txt: Added.
- css3/scroll-snap/scroll-snap-right-to-left-initial-snapping.html: Added.
- 3:15 AM Changeset in webkit [278867] by
-
- 2 edits in trunk/Tools
[GTK] Unreviewed test gardening. Gardened three API GTK test failures.
These tests were constant failures in EWS and were slowing down the
bot. The tests are the following:
- /WebKit2Gtk/TestContextMenu:/webkit/WebKitWebPage/context-menu.
- /WebKit2Gtk/TestContextMenu:/webkit/WebKitWebPage/context-menu-node.
- /TestWTF:WTF_DataMutex.DoubleLockDeathTest.
- TestWebKitAPI/glib/TestExpectations.json:
- 2:58 AM Changeset in webkit [278866] by
-
- 3 edits in trunk/LayoutTests
[GStreamer] media/video-seek-after-end.html is no longer flakey
https://bugs.webkit.org/show_bug.cgi?id=203079
<rdar://problem/56366173>
Reviewed by Philippe Normand.
Increased test timeout to something higher than 1300ms to prevent the test failure.
- media/video-seek-after-end.html: Increased timeout.
- platform/glib/TestExpectations: Unskipped the test.
- 2:19 AM Changeset in webkit [278865] by
-
- 5 edits in trunk/Source/WebCore
[css-flexbox] Do not compute the min-max sizes of flex items twice
https://bugs.webkit.org/show_bug.cgi?id=226463
Reviewed by Simon Fraser.
When determining the flex base size, the item’s min and max main sizes are ignored (no clamping occurs).
Those limits are used to compute the item's hypothetical main size and also later when the flexible
lengths are resolved. The thing is that we were running the code that clamps the flex item size twice instead
of computing those limits once and apply them twice.
From now one, we just compute them once and store the limits in a std::pair in the FlexItem class. This means
that the FlexItem is able to compute the hypothetical main size on its own and does not need it to be passed
as an argument.
No new tests as this is already being tested by dozens of tests.
- rendering/FlexibleBoxAlgorithm.cpp:
(WebCore::FlexItem::FlexItem):
(WebCore::FlexItem::constrainSizeByMinMax const): Clamp the passed in size by the stored min & max sizes.
- rendering/FlexibleBoxAlgorithm.h:
- rendering/RenderFlexibleBox.cpp:
(WebCore::RenderFlexibleBox::computeFlexItemMinMaxSizes): Renamed from adjustChildSizeForMinAndMax and
without the childSize argument which is no longer needed.
(WebCore::RenderFlexibleBox::constructFlexItem): Use constrainSizeByMinMax.
(WebCore::RenderFlexibleBox::resolveFlexibleLengths): Ditto.
(WebCore::RenderFlexibleBox::adjustChildSizeForMinAndMax): Deleted.
- rendering/RenderFlexibleBox.h:
- 1:32 AM Changeset in webkit [278864] by
-
- 4 edits2 moves2 adds in trunk
logged in GitHub issue pages have bad layout for "Notifications Customize" link
https://bugs.webkit.org/show_bug.cgi?id=226859
Reviewed by Alan Bujtas.
LayoutTests/imported/w3c:
- web-platform-tests/html/rendering/the-details-element/summary-display-list-item-002-expected.html: Added.
- web-platform-tests/html/rendering/the-details-element/summary-display-list-item-002.html: Added.
Source/WebCore:
Test: imported/w3c/web-platform-tests/html/rendering/the-details-element/summary-display-list-item-002.html
In r278280 we made display:list-item on <summary> elements to fallback to display:flexbox in order not to
show two markers. However there is no reason why it should be a flexible box instead of a block container,
actually is causing issues in some sites. We better fallback to block instead of flexbox renderer.
- rendering/RenderElement.cpp:
(WebCore::RenderElement::createFor): create a RenderBlockFlow when OnlyCreateBlockAndFlexboxRenderers is
specified (right now only by summary elements).
- 12:40 AM Changeset in webkit [278863] by
-
- 6 edits in trunk/Source/WebCore
[CG] Create a sub-image when drawing an image only if the destination image will be enlarged, rotated or sheared
https://bugs.webkit.org/show_bug.cgi?id=226916
<rdar://71712144>
Reviewed by Simon Fraser.
When drawing only part of the image, we may want to create a sub-image
first and draw this sub-image. But creating the sub-image is expensive.
So we use SubimageCacheWithTimer to avoid recreating them very often.
If the same part of the image is drawn multiple times, creating the sub-
image is a win. But if the part of the image is drawn only once, drawing
the whole image to a clipped context will be faster. In some cases, we
must to draw the part of the image through a sub-image. For example, if
the destination rectangle stretches only a part of an image, interpolating
the sub-image will give the correct display.
In this patch GraphicsContextCG::drawNativeImage() is re-factored to do
the following:
1) Normalize the srcRect and destRect
2) Simplify calculation of adjustDestRect and subimageRect using FloatRect
and FloatSize math
3) Move the heuristic which decides whether to use a sub-image or not to
a separate function.
A sub-image will be created if:
1) An interpolation will be performed when drawing the sub-image
2) and one of the following:
a) If the context is rotated or sheared
b) If destRect.size() / srcRect.size() is non-uniformly scaled
c) If destRect.size() / srcRect.size() is uniformly enlarging in
both direction
- html/canvas/CanvasRenderingContext2DBase.cpp:
(WebCore::normalizeRect): Deleted.
- platform/graphics/FloatRect.cpp:
(WebCore::normalizeRect):
- platform/graphics/FloatRect.h:
- platform/graphics/cg/GraphicsContextCG.cpp:
(WebCore::shouldUseSubimage):
(WebCore::getSubimage):
(WebCore::imageLogicalSize):
(WebCore::GraphicsContextCG::drawNativeImage):
- platform/graphics/transforms/AffineTransform.h:
(WebCore::AffineTransform::isRotateOrShear const):
- 12:08 AM Changeset in webkit [278862] by
-
- 17 edits in trunk
[css-scroll-snap] New snap containers always snap to the first scroll position
https://bugs.webkit.org/show_bug.cgi?id=226630
Reviewed by Simon Fraser.
Source/WebCore:
There are two situations where we should not immediately snap to the first snap position
of a scroller after initial layout:
- If that scroll is right-to-left. In this case the last snap position is closest to the origin.
- If the scroller uses proximity snapping and the first snap position is far enough away from the origin that it isn't yet active.
Previously, WebKit was always snapping to the first position. The change fixes that by
not snapping to the 0 index snap point immediately after initial layout and only snapping
to an eligible snap positions after running a snap point search.
No new tests. This change fixes two existing WPT tests:
- imported/w3c/web-platform-tests/css/css-scroll-snap/scroll-target-snap-003.htm
- imported/w3c/web-platform-tests/css/css-scroll-snap/snap-after-initial-layout/scroll-snap-initial-layout-000.htm
- page/scrolling/ScrollSnapOffsetsInfo.cpp:
(WebCore::closestSnapOffsetWithInfoAndAxis): Pull in isNearEnoughToOffsetForProximity as
an anonymous function and use it to avoid snapping to the first and last position if
they are too far for proximity snapping.
- page/scrolling/ScrollingStateScrollingNode.cpp:
(WebCore::ScrollingStateScrollingNode::dumpProperties const): Use invalidSnapOffsetIndex as
the default for the snap index property. Interpret 0 as a valid snap position.
- page/scrolling/ScrollingStateScrollingNode.h: Ditto.
- page/scrolling/ScrollingTreeScrollingNode.cpp:
(WebCore::ScrollingTreeScrollingNode::dumpProperties const): Ditto.
- page/scrolling/ScrollingTreeScrollingNode.h: Ditto.
- platform/ScrollAnimator.cpp:
(WebCore::ScrollAnimator::resnapAfterLayout): Added this method that passes through to ScrollController.
(WebCore::ScrollAnimator::updateActiveScrollSnapIndexForOffset): Simplified setActiveScrollSnapIndicesForOffset
into updateActiveScrollSnapIndexForClientOffset.
- platform/ScrollAnimator.h: Added method definition.
- platform/ScrollController.cpp:
(WebCore::ScrollController::setSnapOffsetsInfo): Use updateActiveScrollSnapIndexForClientOffset now.
(WebCore::ScrollController::activeScrollSnapIndexForAxis const): Return invalidSnapOffsetIndex
when snapping is disabled.
(WebCore::ScrollController::setNearestScrollSnapIndexForAxisAndOffset): No longer clamp the scroll
position to the first and last snap points. We might be scrolling to a position before or after
them that isn't subject to proximity snapping.
(WebCore::ScrollController::updateActiveScrollSnapIndexForClientOffset): Renamed from setActiveScrollSnapIndicesForOffset
because it always just took the client's current offset.
(WebCore::ScrollController::resnapAfterLayout): Added this helper which snaps in axes that aren't currently
snapped to a snap position after a layout.
- platform/ScrollController.h: Removed unused method that used 0 incorrectly as an invalid snap position.
Update method definitions.
- platform/ScrollSnapAnimatorState.h: Use invalidSnapOffsetIndex to signify no snapping.
- platform/ScrollSnapAnimatorState.cpp:
(WebCore::ScrollSnapAnimatorState::targetOffsetForStartOffset const): No longer clamp offsets to first and
last scroll snap offsets.
- platform/ScrollableArea.cpp:
(WebCore::ScrollableArea::currentHorizontalSnapPointIndex const): Ditto.
(WebCore::ScrollableArea::currentVerticalSnapPointIndex const): Ditto.
(WebCore::ScrollableArea::resnapAfterLayout): Call into ScrollController::resnapAfterLayout.
Source/WebKit:
- UIProcess/RemoteLayerTree/RemoteScrollingCoordinatorProxy.h: Use invalidSnapOffsetIndex
instead of 0 to initialize the snap position.
LayoutTests:
- TestExpectations: Mark two WPT tests as passing.
Jun 14, 2021:
- 11:46 PM Changeset in webkit [278861] by
-
- 4 edits17 adds in trunk/LayoutTests
Import WPT webrtc-extensions
https://bugs.webkit.org/show_bug.cgi?id=226961
Reviewed by Eric Carlson.
LayoutTests/imported/w3c:
- resources/import-expectations.json:
- web-platform-tests/webrtc-extensions/: Added.
LayoutTests:
- tests-options.json:
- 10:26 PM Changeset in webkit [278860] by
-
- 6 edits in trunk/LayoutTests
[GStreamer] media/track/audio/audio-track-mkv-vorbis-language.html and media/track/video/video-track-mkv-theora-language.html are failing
https://bugs.webkit.org/show_bug.cgi?id=226864
Reviewed by Eric Carlson.
These tests were failing due to an apparently incorrect order of
languages in audio and video tracks.
Examining the video container with tools like mediasource or VLC shows
that the order expected by GStreamer ports is correct. To make these
tests pass I changed the tests to make them language order independent
by fetching audio or video tracks by id.
- media/in-band-tracks.js:
(canplaythrough):
(testAttribute): values' can be also an Object. In that case, fetch tracks by 'id'.
- media/track/audio/audio-track-mkv-vorbis-language-expected.txt: Updated.
- media/track/audio/audio-track-mkv-vorbis-language.html: Pass language expected results as a map indexed by track id.
- media/track/video/video-track-mkv-theora-language-expected.txt: Updated.
- media/track/video/video-track-mkv-theora-language.html: Pass language expected results as a map indexed by track id.
- 9:45 PM Changeset in webkit [278859] by
-
- 3 edits1 add in trunk
[JSC] Workaround ICU uloc_addLikelySubtags / uloc_minimizeSubtags bugs
https://bugs.webkit.org/show_bug.cgi?id=226996
rdar://79250513
Reviewed by Ross Kirsling.
JSTests:
- stress/intl-long-locale-id-maximize-minimize.js: Added.
(shouldBe):
(throw.new.Error):
Source/JavaScriptCore:
ICU has bugs that uloc_addLikelySubtags / uloc_minimizeSubtags cannot handle very long locale ID that exceeds ULOC_FULLNAME_CAPACITY,
while these functions can take arbitrary sized buffer for output. This can be achieved simply by (1) attaching many unicode extensions,
or (2) having many variants.
In this patch, we add a workaround: if uloc_addLikelySubtags / uloc_minimizeSubtags failed, we perform them without having locale ID
keywords part. After performing the operations, we append these keywords back.
This is workaround, and still this workaround is not complete since we could have many variants. In that case, uloc_addLikelySubtags / uloc_minimizeSubtags
still fails, and in that case, for now, we give up performing uloc_addLikelySubtags / uloc_minimizeSubtags. Fixing this needs to be
done in ICU side: https://unicode-org.atlassian.net/browse/ICU-21639
- runtime/IntlLocale.cpp:
(JSC::IntlLocale::keywordValue const):
(JSC::IntlLocale::maximal):
(JSC::IntlLocale::minimal):
(JSC::IntlLocale::baseName):
- 8:34 PM Changeset in webkit [278858] by
-
- 2 edits in trunk/Source/WebCore
Font::applyTransforms() can redirect GlyphBuffer's internal pointers
https://bugs.webkit.org/show_bug.cgi?id=226997
<rdar://problem/78704666>
Reviewed by Ryosuke Niwa.
After calling Font::applyTransforms(), we need to update any pointers which point into the GlyphBuffer.
- platform/graphics/WidthIterator.cpp:
(WebCore::WidthIterator::applyFontTransforms):
- 6:07 PM Changeset in webkit [278857] by
-
- 1 copy in tags/Safari-612.1.18.1.1
Tag Safari-612.1.18.1.1.
- 6:04 PM Changeset in webkit [278856] by
-
- 7 edits in trunk/Source/JavaScriptCore
Add ldp and stp support for FP registers, plus some bug fixes.
https://bugs.webkit.org/show_bug.cgi?id=226998
rdar://79313717
Reviewed by Robin Morisset.
This patch does the following:
- Add ldp and stp support for FP registers. This simply entails providing wrappers that take FPRegisterID and passing true for the V bit to the underlying loadStoreRegisterPairXXX encoding function. V is for vector (aka floating point). This will cause bit 26 in the instruction to be set indicating that it's loading / storing floating point registers.
- Add ARM64 disassembler support ldp and stp for FP registers. This includes fixing A64DOpcodeLoadStoreRegisterPair::mask to not exclude the FP versions of the instructions.
- Add ARM64Assembler query methods for determining if an immediate is encodable as the signed 12 bit immediate of ldp and stp instructions.
- Fix ldp and stp offset form to take an int instead of an unsigned. The immediate it takes is a 12-bit signed int, not unsigned.
- In loadStoreRegisterPairXXX encoding functions used by the forms of ldp and stp, RELEASE_ASSERT that the passed in immediate is encodable. Unlike ldur / stur, there is no form of ldp / stp that takes the offset in a register that can be used as a fail over. Hence, if the immediate is not encodable, this is a non-recoverable event. The client is responsible for ensuring that the offset is encodable.
- Added some testmasm tests for testing the offset form (as opposed to PreIndex and PostIndex forms) of ldp and stp. We currently only use the offset form in our JITs.
- assembler/ARM64Assembler.h:
(JSC::ARM64Assembler::isValidLDPImm):
(JSC::ARM64Assembler::isValidLDPFPImm):
(JSC::ARM64Assembler::ldp):
(JSC::ARM64Assembler::ldnp):
(JSC::ARM64Assembler::isValidSTPImm):
(JSC::ARM64Assembler::isValidSTPFPImm):
(JSC::ARM64Assembler::stp):
(JSC::ARM64Assembler::stnp):
(JSC::ARM64Assembler::loadStoreRegisterPairPostIndex):
(JSC::ARM64Assembler::loadStoreRegisterPairPreIndex):
(JSC::ARM64Assembler::loadStoreRegisterPairOffset):
(JSC::ARM64Assembler::loadStoreRegisterPairNonTemporal):
- assembler/AssemblerCommon.h:
(JSC::isValidSignedImm7):
- assembler/MacroAssemblerARM64.h:
(JSC::MacroAssemblerARM64::loadPair64):
(JSC::MacroAssemblerARM64::storePair64):
- assembler/testmasm.cpp:
(JSC::testLoadStorePair64Int64):
(JSC::testLoadStorePair64Double):
- disassembler/ARM64/A64DOpcode.cpp:
(JSC::ARM64Disassembler::A64DOpcodeLoadStoreRegisterPair::format):
- disassembler/ARM64/A64DOpcode.h:
- 6:02 PM Changeset in webkit [278855] by
-
- 1 copy in tags/Safari-612.1.18.0.1
Tag Safari-612.1.18.0.1.
- 5:54 PM Changeset in webkit [278854] by
-
- 2 edits in trunk/Tools
run-javascriptcore-tests should print output when a test binary fails by default
https://bugs.webkit.org/show_bug.cgi?id=226985
Reviewed by Mark Lam.
- Scripts/run-javascriptcore-tests:
(runTest):
- 5:46 PM Changeset in webkit [278853] by
-
- 2 edits in trunk/Source/WebInspectorUI
[REGRESSION: r276616] Uncaught Exception: TypeError: undefined is not an object (evaluating 'this._setupCodeMirror.getValue')
https://bugs.webkit.org/show_bug.cgi?id=226995
Reviewed by Devin Rousso.
Audit test groups share a single setup editor for all their test cases, leaving individual test cases without
their own setup editor. In this case, we should not attempt to save the setup script for every test, and instead
allow the instance ofAuditTestGroupContentViewthat does have a setup editor to save the script.
- UserInterface/Views/AuditTestContentView.js:
(WI.AuditTestContentView.prototype.saveEditedData):
- 4:40 PM Changeset in webkit [278852] by
-
- 4 edits in trunk/Source/WebCore
Remove Storage Access API quirk on live.com
https://bugs.webkit.org/show_bug.cgi?id=226990
<rdar://problem/79218986>
Reviewed by Brent Fulgham.
The quirk on live.com is unnecessary, and we should remove it.
- dom/DocumentStorageAccess.cpp:
(WebCore::DocumentStorageAccess::requestStorageAccessQuirk):
- page/Quirks.cpp:
(WebCore::isStorageAccessQuirkDomainAndElement):
(WebCore::Quirks::requestStorageAccessAndHandleClick const):
(WebCore::Quirks::mapToTopDomain): Deleted.
- page/Quirks.h:
- 4:39 PM Changeset in webkit [278851] by
-
- 5 edits3 deletes in trunk
Unreviewed, reverting r278842.
introduced a failing test
Reverted changeset:
"Fix RTCDataChannelInit::decode"
https://bugs.webkit.org/show_bug.cgi?id=226968
https://commits.webkit.org/r278842
- 3:04 PM Changeset in webkit [278850] by
-
- 2 edits in trunk/Source/WebCore
[macOS] TouchBar playback speed controls don't work
https://bugs.webkit.org/show_bug.cgi?id=226987
<rdar://problem/79216098>
Reviewed by Eric Carlson.
Override
setRate:andsetDefaultPlaybackRate:instead of just having an ivar so that
TouchBar playback speed controls actually affect the corresponding<video>.
- platform/mac/WebPlaybackControlsManager.mm:
(-[WebPlaybackControlsManager defaultPlaybackRate]): Added.
(-[WebPlaybackControlsManager setDefaultPlaybackRate:]): Added.
(-[WebPlaybackControlsManager rate]): Added.
(-[WebPlaybackControlsManager setRate:]): Added.
- 2:36 PM Changeset in webkit [278849] by
-
- 55 edits in trunk/Source
Pass PAL::SessionID by value instead of reference
https://bugs.webkit.org/show_bug.cgi?id=226983
Patch by Alex Christensen <achristensen@webkit.org> on 2021-06-14
Reviewed by Geoffrey Garen.
Source/WebCore:
It's a wrapper around an integer which will pass in a register.
There's no reason to pass a pointer to it.
- loader/CrossOriginAccessControl.cpp:
(WebCore::validatePreflightResponse):
- loader/CrossOriginAccessControl.h:
- loader/CrossOriginPreflightResultCache.cpp:
(WebCore::CrossOriginPreflightResultCache::appendEntry):
(WebCore::CrossOriginPreflightResultCache::canSkipPreflight):
- loader/CrossOriginPreflightResultCache.h:
- loader/EmptyClients.cpp:
- loader/cache/CachedApplicationManifest.cpp:
(WebCore::CachedApplicationManifest::CachedApplicationManifest):
- loader/cache/CachedApplicationManifest.h:
- loader/cache/CachedCSSStyleSheet.cpp:
(WebCore::CachedCSSStyleSheet::CachedCSSStyleSheet):
- loader/cache/CachedCSSStyleSheet.h:
- loader/cache/CachedFont.cpp:
(WebCore::CachedFont::CachedFont):
- loader/cache/CachedFont.h:
- loader/cache/CachedImage.cpp:
(WebCore::CachedImage::CachedImage):
- loader/cache/CachedImage.h:
- loader/cache/CachedRawResource.cpp:
(WebCore::CachedRawResource::CachedRawResource):
- loader/cache/CachedRawResource.h:
- loader/cache/CachedResource.cpp:
(WebCore::CachedResource::CachedResource):
- loader/cache/CachedResource.h:
- loader/cache/CachedResourceLoader.cpp:
(WebCore::createResource):
(WebCore::CachedResourceLoader::updateCachedResourceWithCurrentRequest):
- loader/cache/CachedResourceLoader.h:
- loader/cache/CachedSVGDocument.cpp:
(WebCore::CachedSVGDocument::CachedSVGDocument):
- loader/cache/CachedSVGDocument.h:
- loader/cache/CachedSVGFont.cpp:
(WebCore::CachedSVGFont::CachedSVGFont):
- loader/cache/CachedSVGFont.h:
- loader/cache/CachedScript.cpp:
(WebCore::CachedScript::CachedScript):
- loader/cache/CachedScript.h:
- loader/cache/CachedTextTrack.cpp:
(WebCore::CachedTextTrack::CachedTextTrack):
- loader/cache/CachedTextTrack.h:
- loader/cache/CachedXSLStyleSheet.cpp:
(WebCore::CachedXSLStyleSheet::CachedXSLStyleSheet):
- loader/cache/CachedXSLStyleSheet.h:
- page/DatabaseProvider.h:
- platform/WebCoreCrossThreadCopier.cpp:
(WTF::PAL::SessionID>::copy):
- platform/WebCoreCrossThreadCopier.h:
- storage/StorageNamespaceProvider.cpp:
(WebCore::StorageNamespaceProvider::setSessionIDForTesting):
- storage/StorageNamespaceProvider.h:
Source/WebKit:
- NetworkProcess/NetworkProcess.cpp:
(WebKit::NetworkProcess::newTestingSession):
(WebKit::NetworkProcess::ensureSession):
(WebKit::NetworkProcess::storageSession const):
(WebKit::NetworkProcess::findCacheEngine):
(WebKit::NetworkProcess::ensureCacheEngine):
(WebKit::NetworkProcess::removeCacheEngine):
(WebKit::NetworkProcess::flushCookies):
(WebKit::NetworkProcess::platformFlushCookies):
(WebKit::NetworkProcess::addWebPageNetworkParameters):
(WebKit::NetworkProcess::removeWebPageNetworkParameters):
(WebKit::NetworkProcess::countNonDefaultSessionSets):
- NetworkProcess/NetworkProcess.h:
- NetworkProcess/NetworkResourceLoader.cpp:
(WebKit::NetworkResourceLoader::shouldLogCookieInformation):
- NetworkProcess/NetworkResourceLoader.h:
- NetworkProcess/NetworkStorageSessionProvider.h:
- NetworkProcess/cache/NetworkCache.cpp:
(WebKit::NetworkCache::makeUseDecision):
- NetworkProcess/cache/NetworkCache.h:
(WebKit::NetworkCache::Cache::sessionID const):
- NetworkProcess/cocoa/NetworkProcessCocoa.mm:
(WebKit::NetworkProcess::flushCookies):
(WebKit::NetworkProcess::platformFlushCookies):
- NetworkProcess/cocoa/NetworkSessionCocoa.mm:
(WebKit::configurationForSessionID):
- UIProcess/Network/NetworkProcessProxy.cpp:
(WebKit::NetworkProcessProxy::flushCookies):
- UIProcess/Network/NetworkProcessProxy.h:
- UIProcess/WebProcessPool.h:
- WebProcess/Databases/WebDatabaseProvider.cpp:
(WebKit::WebDatabaseProvider::idbConnectionToServerForSession):
- WebProcess/Databases/WebDatabaseProvider.h:
Source/WebKitLegacy:
- Storage/WebDatabaseProvider.cpp:
(WebDatabaseProvider::idbConnectionToServerForSession):
- Storage/WebDatabaseProvider.h:
- WebCoreSupport/NetworkStorageSessionMap.cpp:
(NetworkStorageSessionMap::storageSession):
(NetworkStorageSessionMap::ensureSession):
(NetworkStorageSessionMap::destroySession):
- WebCoreSupport/NetworkStorageSessionMap.h:
- 1:01 PM Changeset in webkit [278848] by
-
- 5 edits in trunk
Web Inspector: CSS variables not handled as case sensitive
https://bugs.webkit.org/show_bug.cgi?id=226875
Reviewed by Devin Rousso.
Source/WebCore:
Test: inspector/css/overridden-property.html
CSS variables support distinct declarations with only differences in cases. Previously, we naively converted all
property names to lowercase, instead of properly providing variable names in their original case.
- inspector/InspectorStyleSheet.cpp:
(WebCore::InspectorStyle::styleWithProperties const):
LayoutTests:
Add a test to make sure that CSS variable declarations that vary only in case do not override each other.
- inspector/css/overridden-property-expected.txt:
- inspector/css/overridden-property.html:
- 12:54 PM Changeset in webkit [278847] by
-
- 3 edits in trunk/Tools
Bring up an Apple-BigSur-AppleSilicon-Release-Test262-Tests queue
https://bugs.webkit.org/show_bug.cgi?id=226602
Reviewed by Aakash Jain.
- CISupport/build-webkit-org/config.json:
- CISupport/build-webkit-org/public_html/dashboard/Scripts/WebKitBuildbot.js:
(WebKitBuildbot):
- 12:44 PM Changeset in webkit [278846] by
-
- 8 edits in trunk/Source/JavaScriptCore
Add Air opcode sub32/64(Reg, Imm, Reg) form for ARM64 and select this instruction in Air
https://bugs.webkit.org/show_bug.cgi?id=226937
Patch by Yijia Huang <Yijia Huang> on 2021-06-14
Reviewed by Saam Barati.
Previously, Air arm64 sub32/64 utilize sub(Imm, Tmp) at optlevel = 0 and
add(Tmp, -Imm) at optlevel > 0 to perform and optimize sub(Tmp, Imm, Tmp).
The issue with this is that we were not eliding redundant operations.
For example:
B3 IR
@0 = Trunc(ArgumentReg(0))
@1 = Const
@2 = Sub(@0, @1)
@3 = Return(@2)
Old optimized Air IR
OptLevel = 0
Move %x0, %tmp1, @0
Move $Const, %tmp2, @1
Move %tmp1, %tmp0, @2 Redundant
Sub $Const, %tmp0, @2
Move %tmp0, %x0, @3
Ret32 %x0, @3
To remove those redundant instructions, Air arm64 sub32/64 opcode should
indicate a new form sub(Tmp, Imm, Tmp).
New optimized Air IR
OptLevel = 0
Move %x0, %tmp1, @0
Move $Const, %tmp2, @1
Sub %tmp1, $Const, %tmp0, @2
Move %tmp0, %x0, @3
Ret32 %x0, @3
- assembler/MacroAssemblerARM64.h:
(JSC::MacroAssemblerARM64::sub32):
(JSC::MacroAssemblerARM64::sub64):
- assembler/testmasm.cpp:
(JSC::testSub32Args):
(JSC::testSub32Imm):
(JSC::testSub32ArgImm):
(JSC::testSub64Imm32):
(JSC::testSub64ArgImm32):
(JSC::testSub64Imm64):
(JSC::testSub64ArgImm64):
- b3/B3ReduceStrength.cpp:
- b3/air/AirOpcode.opcodes:
- b3/testb3.h:
- b3/testb3_2.cpp:
(testSubArgs32ZeroExtend):
- b3/testb3_3.cpp:
(addArgTests):
- 12:20 PM Changeset in webkit [278845] by
-
- 2 edits in branches/safari-611-branch/Source/WebKit
Apply patch. rdar://problem/77619702
- 11:44 AM Changeset in webkit [278844] by
-
- 8 edits in trunk/Source
Versioning.
WebKit-7612.1.20
- 11:41 AM Changeset in webkit [278843] by
-
- 8 edits in branches/safari-612.1.19-branch/Source
Versioning.
WebKit-7612.1.19
- 11:40 AM Changeset in webkit [278842] by
-
- 5 edits3 adds in trunk
Fix RTCDataChannelInit::decode
https://bugs.webkit.org/show_bug.cgi?id=226968
Reviewed by Eric Carlson.
LayoutTests/imported/w3c:
- web-platform-tests/webrtc-extensions/transfer-datachannel-service-worker.https-expected.txt: Added.
- web-platform-tests/webrtc-extensions/transfer-datachannel-service-worker.https.html: Added.
- web-platform-tests/webrtc-extensions/transfer-datachannel-service-worker.js: Added.
(onmessage):
Source/WebCore:
Test: imported/w3c/web-platform-tests/webrtc-extensions/transfer-datachannel-service-worker.https.html
- platform/mediastream/RTCDataChannelHandler.h:
(WebCore::RTCDataChannelInit::decode):
Make sure to use optionals of optionals.
LayoutTests:
- platform/mac-wk1/TestExpectations:
- 11:39 AM Changeset in webkit [278841] by
-
- 2 edits in trunk/Source/WebKit
Accessing navigator.mediaDevices.enumerateDevices after granting permission
https://bugs.webkit.org/show_bug.cgi?id=226969
<rdar://problem/79286449>
Reviewed by Eric Carlson.
- UIProcess/Cocoa/MediaPermissionUtilities.mm:
(WebKit::checkUsageDescriptionStringForType):
Check for NSCameraUsageDescription instead of NSMicrophoneUsageDescription for camera.
- 11:37 AM Changeset in webkit [278840] by
-
- 1 edit3 moves1 add in trunk/LayoutTests/imported/w3c
Move Transferable RTCDataChannel tests to WPT
https://bugs.webkit.org/show_bug.cgi?id=226967
Reviewed by Eric Carlson.
- web-platform-tests/webrtc-extensions/transfer-datachannel-expected.txt: Renamed from LayoutTests/http/wpt/webrtc/datachannel-worker-expected.txt.
- web-platform-tests/webrtc-extensions/transfer-datachannel-worker.js: Renamed from LayoutTests/http/wpt/webrtc/datachannel-worker.js.
- web-platform-tests/webrtc-extensions/transfer-datachannel.html: Renamed from LayoutTests/http/wpt/webrtc/datachannel-worker.html.
- 11:27 AM Changeset in webkit [278839] by
-
- 1 copy in branches/safari-612.1.19-branch
New branch.
- 11:20 AM Changeset in webkit [278838] by
-
- 2 edits in trunk/Source/WebCore
[AppleWin] Fix build failure
https://bugs.webkit.org/show_bug.cgi?id=226966
<rdar://79228536>
Reviewed by Eric Carlson.
Disable warning causing the build failure.
- platform/cf/MediaAccessibilitySoftLink.h:
- 10:47 AM Changeset in webkit [278837] by
-
- 2 edits in trunk/Source/WebCore
Avoid converting SecurityOrigins to Strings in SecurityPolicy
https://bugs.webkit.org/show_bug.cgi?id=226976
Reviewed by Geoffrey Garen.
Avoid converting SecurityOrigins to Strings in SecurityPolicy, by using a SecurityOriginData
as key in the originAccessMap instead of a String. Getting a SecurityOriginData from a
SecurityOrigin is very cheap.
- page/SecurityPolicy.cpp:
(WebCore::SecurityPolicy::isAccessAllowed):
(WebCore::SecurityPolicy::addOriginAccessAllowlistEntry):
(WebCore::SecurityPolicy::removeOriginAccessAllowlistEntry):
- 10:41 AM Changeset in webkit [278836] by
-
- 8 edits in branches/safari-612.1.18.1-branch/Source
Versioning.
WebKit-7612.1.18.1.1
- 10:40 AM Changeset in webkit [278835] by
-
- 8 edits in branches/safari-612.1.18.0-branch/Source
Versioning.
WebKit-7612.1.18.0.1
- 10:34 AM Changeset in webkit [278834] by
-
- 5 edits in trunk/Source/WebCore
[Monterey] Support building WebKit (Part 2)
https://bugs.webkit.org/show_bug.cgi?id=226846
<rdar://problem/79095148>
Reviewed by Chris Dumez.
Source/WebCore:
Covered by exisiting tests.
- html/HTMLElement.cpp:
(WebCore::HTMLElement::updateWithTextRecognitionResult): Explicitly define second arugment
to avoid 'missing field' compilation error.
Source/WebCore/PAL:
- pal/spi/cf/CFNetworkSPI.h: Add nw_connection_privacy_stance_t.
- pal/spi/cocoa/PassKitSPI.h: Add PKPaymentCoupon.
- 10:31 AM Changeset in webkit [278833] by
-
- 2 edits in trunk/Source/WebCore
Micro-optimize ResourceResponseBase::sanitizeHTTPHeaderFields()
https://bugs.webkit.org/show_bug.cgi?id=226977
Reviewed by Geoffrey Garen.
Micro-optimize ResourceResponseBase::sanitizeHTTPHeaderFields() as it is not as efficient as it could be and it
shows on profiles. In particular, rely on Vector::removeAllMatching() more instead of reconstructing HTTPHeaderMaps
& Vectors.
- platform/network/ResourceResponseBase.cpp:
(WebCore::ResourceResponseBase::sanitizeHTTPHeaderFieldsAccordingToTainting):
(WebCore::ResourceResponseBase::sanitizeHTTPHeaderFields):
- 10:27 AM Changeset in webkit [278832] by
-
- 7 edits1 add in trunk
AcceleratedImageBuffer not instantiated but objects are punned to the type
https://bugs.webkit.org/show_bug.cgi?id=226917
Patch by Kimmo Kinnunen <kkinnunen@apple.com> on 2021-06-14
Reviewed by Said Abou-Hallawa.
Source/WebCore:
AcceleratedImageBuffer::create would create instance
of ConcreteImageBuffer, not AcceleratedImageBuffer.
Fix by adding a correct create static function.
Tested by new test:
TestWebKitAPI.ImageBufferTests.ImageBufferSubTypeCreateCreatesSubtypes
- platform/graphics/PlatformImageBuffer.h:
(WebCore::IOSurfaceImageBuffer::create):
(WebCore::IOSurfaceImageBuffer::IOSurfaceImageBuffer):
Add the correct factory functions and public constructors.
- platform/graphics/cg/ImageBufferCGBitmapBackend.h:
Add export statements that are needed for the added test.
- platform/graphics/coreimage/FilterEffectRendererCoreImage.mm:
(WebCore::FilterEffectRendererCoreImage::imageForSourceGraphic):
(WebCore::FilterEffectRendererCoreImage::renderToImageBuffer):
Change the image creation site to create the explicit IOSurfaceImageBuffer
since that's what it's trying to create.
Change the type-punning call sites to use explicit
IOSurfaceImageBuffer name, as that's what the call site
is intending to use.
Tools:
Add new test
TestWebKitAPI.ImageBufferTests.ImageBufferSubTypeCreateCreatesSubtypes
to test that the code compiles.
Before, it would fail with:
`error: cannot initialize a member subobject of type 'typename PtrTraits::StorageType' (aka 'WebCore::AcceleratedImageBuffer *') with an
rvalue of type 'WebCore::ConcreteImageBuffer<WebCore::ImageBufferIOSurfaceBackend> *'`
- TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
- TestWebKitAPI/Tests/WebCore/ImageBufferTests.cpp: Added.
(TestWebKitAPI::TEST):
- 10:19 AM Changeset in webkit [278831] by
-
- 2 edits in trunk/Source/WebKit
[iOS] Tighten the WebAuthn Sandbox
https://bugs.webkit.org/show_bug.cgi?id=226944
<rdar://problem/74721792>
Reviewed by Per Arne Vollan.
Revise the WebAuthn process sandbox to remove the many things originally copied from
the WebContent sandbox that are not needed.
- Resources/SandboxProfiles/ios/com.apple.WebKit.WebAuthn.sb:
- 9:14 AM Changeset in webkit [278830] by
-
- 6 edits in trunk/Source/JavaScriptCore
Unreviewed, reverting r278734.
https://bugs.webkit.org/show_bug.cgi?id=226973
broke jsc testapi
Reverted changeset:
"Shouldn't drain the micro task queue when calling out to
ObjC"
https://bugs.webkit.org/show_bug.cgi?id=161942
https://trac.webkit.org/changeset/278734
- 8:39 AM Changeset in webkit [278829] by
-
- 1 edit3 deletes in trunk/LayoutTests
[LayoutTests] Delete unused LayoutTests/plugins resources
https://bugs.webkit.org/show_bug.cgi?id=226837
Reviewed by Jonathan Bedard.
- plugins/resources/frame-with-plugin-subframe.html: Removed.
- plugins/resources/lines.swf: Removed.
- plugins/resources/plugin-page.html: Removed.
- 8:37 AM Changeset in webkit [278828] by
-
- 1 copy in branches/safari-612.1.18.1-branch
New branch.
- 8:33 AM Changeset in webkit [278827] by
-
- 1 copy in branches/safari-612.1.18.0-branch
New branch.
- 8:28 AM Changeset in webkit [278826] by
-
- 2 edits in trunk/LayoutTests
[AppleWin] Update test expectations
https://bugs.webkit.org/show_bug.cgi?id=226970
Unreviewed test gardening.
- platform/win/TestExpectations:
- 6:39 AM Changeset in webkit [278825] by
-
- 4 edits4 adds in trunk
[LFC][TFC] Add support for flexing columns when the content is also flexing
https://bugs.webkit.org/show_bug.cgi?id=226959
Source/WebCore:
Reviewed by Antti Koivisto.
This patch adds support for (the very common) cases when the minimum content width < maximum content width
(e.g. text content wraps across multiple lines) and the parent column does not have fixed width either.
Note that in some cases now table layout agrees with Chrome/Firefox and not with trunk WebKit
(see table-space-distribution-simple-mismatching.htnl).
Tests: fast/layoutformattingcontext/table-space-distribution-simple-mismatching.html
fast/layoutformattingcontext/table-space-distribution-simple2.html
- layout/formattingContexts/table/TableLayout.cpp:
(WebCore::Layout::TableFormattingContext::TableLayout::distributedHorizontalSpace):
LayoutTests:
Reviewed by Antti Koivisto.
- TestExpectations: LFC matches Chrome and Firefox.
- fast/layoutformattingcontext/table-space-distribution-simple-mismatching-expected.txt: Added.
- fast/layoutformattingcontext/table-space-distribution-simple-mismatching.html: Added.
- fast/layoutformattingcontext/table-space-distribution-simple2-expected.html: Added.
- fast/layoutformattingcontext/table-space-distribution-simple2.html: Added.
- 2:28 AM Changeset in webkit [278824] by
-
- 4 edits in trunk/Source
[iOS] Sync Accessibility preferences
https://bugs.webkit.org/show_bug.cgi?id=226738
<rdar://77922839>
Reviewed by Brent Fulgham.
Source/WebKit:
Only attempt to modify Accessibility preferences if they have changed.
- Shared/AccessibilityPreferences.cpp: Replaced.
- Shared/AccessibilityPreferences.h: Replaced.
- WebProcess/cocoa/WebProcessCocoa.mm:
(WebKit::WebProcess::accessibilityPreferencesDidChange):
Source/WTF:
Enable Per-App Accessibility preferences on iOS.
- wtf/PlatformHave.h:
- 12:38 AM WebKitGTK/2.32.x edited by
- (diff)
