WebKit-https.git
5 months agoUnreviewed, fix unused variable warnings in release builds
mcatanzaro@igalia.com [Mon, 13 May 2019 15:39:31 +0000 (15:39 +0000)]
Unreviewed, fix unused variable warnings in release builds

Source/WebKit:

* UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::clearWebProcessHasUploads):

Source/WTF:

* wtf/URLHelpers.cpp:
(WTF::URLHelpers::escapeUnsafeCharacters):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245234 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoWHLSLPrepare.cpp always recompiles, even if nothing was changed
darin@apple.com [Mon, 13 May 2019 13:59:59 +0000 (13:59 +0000)]
WHLSLPrepare.cpp always recompiles, even if nothing was changed
https://bugs.webkit.org/show_bug.cgi?id=197151

Reviewed by Dan Bernstein and Keith Rollin.

* DerivedSources-input.xcfilelist: Script updated this automatically after
DerivedSources.make was corrected.
* DerivedSources-output.xcfilelist: Ditto, although I had to manually remove
one bogus leftover reference to WHLSLStandardLibrary.cpp.

* DerivedSources.make: Updated the rule that builds WHSLStandardLibrary.h to
no longer refer to nonexistent WHLSLStandardLibrary.cpp. Because the dependency
was on a file that was never created, the rule to regenerate WHSLStandardLibrary.h
was running on every build, instead of only when one of the dependencies changed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245233 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoREGRESSION (r245208): compositing/shared-backing/sharing-bounds-non-clipping-shared...
simon.fraser@apple.com [Mon, 13 May 2019 04:13:06 +0000 (04:13 +0000)]
REGRESSION (r245208): compositing/shared-backing/sharing-bounds-non-clipping-shared-layer.html asserts
https://bugs.webkit.org/show_bug.cgi?id=197818

Approved by Jon Lee.

Skip this test until a fix for bug 197818 lands.

* platform/ios-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245221 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoWhen the set of backing-sharing layers changes, we need to issue a repaint
simon.fraser@apple.com [Mon, 13 May 2019 03:57:38 +0000 (03:57 +0000)]
When the set of backing-sharing layers changes, we need to issue a repaint
https://bugs.webkit.org/show_bug.cgi?id=197825

Reviewed by Zalan Bujtas.

Source/WebCore:

If the set of layers painting into a shared backing store changes, we need
to repaint that backing store. This happens when scrolling as shared layers
enter the visible area.

Test: compositing/shared-backing/overflow-scroll/repaint-shared-on-scroll.html

* rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::setBackingSharingLayers):

LayoutTests:

* compositing/shared-backing/overflow-scroll/repaint-shared-on-scroll-expected.html: Added.
* compositing/shared-backing/overflow-scroll/repaint-shared-on-scroll.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245220 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoAdd logging for RenderLayer clip rects
simon.fraser@apple.com [Mon, 13 May 2019 03:07:52 +0000 (03:07 +0000)]
Add logging for RenderLayer clip rects
https://bugs.webkit.org/show_bug.cgi?id=197547

Reviewed by Zalan Bujtas.

Add a ClipRects log channel, and stream output for ClipRect and ClipRects.

The ClipRect code is performance sensitive, even in debug, so guard the log sites
with clipRectsLogEnabled() because the macro still evaluates its arguments even if
the channel is disabled (we need some better way to log that doesn't do this).

* platform/Logging.h:
* rendering/ClipRect.cpp:
(WebCore::operator<<):
* rendering/ClipRect.h:
* rendering/RenderLayer.cpp:
(WebCore::operator<<):
(WebCore::RenderLayer::calculateClipRects const):
* rendering/RenderLayer.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245219 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoRefactor composited backing-sharing code
simon.fraser@apple.com [Mon, 13 May 2019 03:01:25 +0000 (03:01 +0000)]
Refactor composited backing-sharing code
https://bugs.webkit.org/show_bug.cgi?id=197824

Reviewed by Zalan Bujtas.

Clean up the backing-sharing code to share more code, and make it easier to understand.

Moves more logic into member functions on BackingSharingState, which are named to make
their functions clearer: startBackingSharingSequence/endBackingSharingSequence.

computeCompositingRequirements() and traverseUnchangedSubtree() now just call
updateBeforeDescendantTraversal/updateAfterDescendantTraversal.

No behavior change.

* rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::willBeDestroyed):
(WebCore::RenderLayerBacking::setBackingSharingLayers): Remove the early return, since
we need to call setBackingProviderLayer() on the sharing layers in both code paths.
(WebCore::RenderLayerBacking::removeBackingSharingLayer):
(WebCore::RenderLayerBacking::clearBackingSharingLayers):
* rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::BackingSharingState::backingProviderCandidate const):
(WebCore::RenderLayerCompositor::BackingSharingState::appendSharingLayer):
(WebCore::RenderLayerCompositor::BackingSharingState::startBackingSharingSequence):
(WebCore::RenderLayerCompositor::BackingSharingState::endBackingSharingSequence):
(WebCore::RenderLayerCompositor::BackingSharingState::updateBeforeDescendantTraversal):
(WebCore::RenderLayerCompositor::BackingSharingState::updateAfterDescendantTraversal):
(WebCore::RenderLayerCompositor::computeCompositingRequirements):
(WebCore::RenderLayerCompositor::traverseUnchangedSubtree):
(WebCore::RenderLayerCompositor::BackingSharingState::resetBackingProviderCandidate): Deleted.
* rendering/RenderLayerCompositor.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245218 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoUse clampTo in AVVideoCaptureSource::setSizeAndFrameRateWithPreset
youenn@apple.com [Mon, 13 May 2019 02:57:03 +0000 (02:57 +0000)]
Use clampTo in AVVideoCaptureSource::setSizeAndFrameRateWithPreset
https://bugs.webkit.org/show_bug.cgi?id=197704

Reviewed by Alex Christensen.

Use clampTo as suggested in bug 196214 review.
No change of behavior.

* platform/mediastream/mac/AVVideoCaptureSource.mm:
(WebCore::AVVideoCaptureSource::setSizeAndFrameRateWithPreset):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245217 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoUse the main screen for screen capture
youenn@apple.com [Mon, 13 May 2019 02:56:13 +0000 (02:56 +0000)]
Use the main screen for screen capture
https://bugs.webkit.org/show_bug.cgi?id=197804
<rdar://problem/47671383>

Reviewed by Eric Carlson.

If the main screen, i.e. the screen that has focus at the time of
selection of the screen to capture, is capturable, add it to the list
of screen devices, but do not add any other screen.
This will make sure the main screen is selected.
Manually tested.

* platform/mediastream/mac/ScreenDisplayCaptureSourceMac.mm:
(WebCore::ScreenDisplayCaptureSourceMac::screenCaptureDevices):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245216 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[Curl] Suppress extra didReceiveAuthenticationChallenge call when accessing a server...
commit-queue@webkit.org [Sun, 12 May 2019 23:05:20 +0000 (23:05 +0000)]
[Curl] Suppress extra didReceiveAuthenticationChallenge call when accessing a server which checks basic auth.
https://bugs.webkit.org/show_bug.cgi?id=197650

Patch by Takashi Komori <Takashi.Komori@sony.com> on 2019-05-12
Reviewed by Fujii Hironori.

Source/WebCore:

When Curl port accesses a page which checks Basic Authentication credential and server trust challenge occurs,
Curl port calls extra didReceiveAuthenticationChallenge unnecessarily.
This is because Curl port discards information about allowed server trust challenge before in NetworkDataTaskCurl::restartWithCredential.

Test: http/tests/ssl/curl/certificate-and-authentication.html

* platform/network/curl/CurlRequest.h:
(WebCore::CurlRequest::isServerTrustEvaluationDisabled):

Source/WebKit:

* NetworkProcess/curl/NetworkDataTaskCurl.cpp:
(WebKit::NetworkDataTaskCurl::restartWithCredential):

LayoutTests:

* TestExpectations:
* http/tests/resources/basic-auth.php: Added.
* http/tests/ssl/curl/certificate-and-authentication-expected.txt: Added.
* http/tests/ssl/curl/certificate-and-authentication.html: Added.
* platform/wincairo-wk1/TestExpectations:
* platform/wincairo/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245215 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[JSC] Compress Watchpoint size by using enum type and Packed<> data structure
ysuzuki@apple.com [Sun, 12 May 2019 22:50:21 +0000 (22:50 +0000)]
[JSC] Compress Watchpoint size by using enum type and Packed<> data structure
https://bugs.webkit.org/show_bug.cgi?id=197730

Reviewed by Filip Pizlo.

Source/JavaScriptCore:

Watchpoint takes 5~ MB memory in Gmail (total memory starts with 400 - 500 MB), so 1~%. Since it is allocated massively,
reducing each size of Watchpoint reduces memory footprint significantly.

As a first step, this patch uses Packed<> and enum to reduce the size of Watchpoint.

1. Watchpoint should have enum type and should not use vtable. vtable takes one pointer, and it is too costly for such a
   memory sensitive objects. We perform downcast and dispatch the method of the derived classes based on this enum. Since
   the # of derived Watchpoint classes are limited (Only 8), we can list up them easily. One unfortunate thing is that
   we cannot do this for destructor so long as we use "delete" for deleting objects. If we dispatch the destructor of derived
   class in the destructor of the base class, we call the destructor of the base class multiple times. delete operator override
   does not help since custom delete operator is called after the destructor is called. While we can fix this issue by always
   using custom deleter, currently we do not since all the watchpoints do not have members which have non trivial destructor.
   Once it is strongly required, we can start using custom deleter, but for now, we do not need to do this.

2. We use Packed<> to compact pointers in Watchpoint. Since Watchpoint is a node of doubly linked list, each one has two
   pointers for prev and next. This is also too costly. PackedPtr reduces the size and makes alignment 1.S

3. We use PackedCellPtr<> for JSCells in Watchpoint. This leverages alignment information and makes pointers smaller in
   Darwin ARM64. One important thing to note here is that since this pointer is packed, it cannot be found by conservative
   GC scan. It is OK for watchpoint since they are allocated in the heap anyway.

We applied this change to Watchpoint and get the following memory reduction. The highlight is that CodeBlockJettisoningWatchpoint in
ARM64 only takes 2 pointers size.

                                                                      ORIGINAL    X86_64   ARM64
    WatchpointSet:                                                    40          32       28
    CodeBlockJettisoningWatchpoint:                                   32          19       15
    StructureStubClearingWatchpoint:                                  56          48       40
    AdaptiveInferredPropertyValueWatchpointBase::StructureWatchpoint: 24          13       11
    AdaptiveInferredPropertyValueWatchpointBase::PropertyWatchpoint:  24          13       11
    FunctionRareData::AllocationProfileClearingWatchpoint:            32          19       15
    ObjectToStringAdaptiveStructureWatchpoint:                        56          48       40
    LLIntPrototypeLoadAdaptiveStructureWatchpoint:                    64          48       48
    DFG::AdaptiveStructureWatchpoint:                                 56          48       40

While we will re-architect the mechanism of Watchpoint, anyway Packed<> mechanism and enum types will be used too.

* CMakeLists.txt:
* JavaScriptCore.xcodeproj/project.pbxproj:
* Sources.txt:
* bytecode/AdaptiveInferredPropertyValueWatchpointBase.h:
* bytecode/CodeBlockJettisoningWatchpoint.h:
* bytecode/CodeOrigin.h:
* bytecode/LLIntPrototypeLoadAdaptiveStructureWatchpoint.cpp:
(JSC::LLIntPrototypeLoadAdaptiveStructureWatchpoint::LLIntPrototypeLoadAdaptiveStructureWatchpoint):
(JSC::LLIntPrototypeLoadAdaptiveStructureWatchpoint::fireInternal):
* bytecode/LLIntPrototypeLoadAdaptiveStructureWatchpoint.h:
* bytecode/StructureStubClearingWatchpoint.cpp:
(JSC::StructureStubClearingWatchpoint::fireInternal):
* bytecode/StructureStubClearingWatchpoint.h:
* bytecode/Watchpoint.cpp:
(JSC::Watchpoint::fire):
* bytecode/Watchpoint.h:
(JSC::Watchpoint::Watchpoint):
* dfg/DFGAdaptiveStructureWatchpoint.cpp:
(JSC::DFG::AdaptiveStructureWatchpoint::AdaptiveStructureWatchpoint):
* dfg/DFGAdaptiveStructureWatchpoint.h:
* heap/PackedCellPtr.h: Added.
* runtime/FunctionRareData.h:
* runtime/ObjectToStringAdaptiveStructureWatchpoint.cpp: Added.
(JSC::ObjectToStringAdaptiveStructureWatchpoint::ObjectToStringAdaptiveStructureWatchpoint):
(JSC::ObjectToStringAdaptiveStructureWatchpoint::install):
(JSC::ObjectToStringAdaptiveStructureWatchpoint::fireInternal):
* runtime/ObjectToStringAdaptiveStructureWatchpoint.h: Added.
* runtime/StructureRareData.cpp:
(JSC::StructureRareData::clearObjectToStringValue):
(JSC::ObjectToStringAdaptiveStructureWatchpoint::ObjectToStringAdaptiveStructureWatchpoint): Deleted.
(JSC::ObjectToStringAdaptiveStructureWatchpoint::install): Deleted.
(JSC::ObjectToStringAdaptiveStructureWatchpoint::fireInternal): Deleted.
* runtime/StructureRareData.h:

Source/WTF:

This patch introduces a new data structures, WTF::Packed, WTF::PackedPtr, and WTF::PackedAlignedPtr.

- WTF::Packed

    WTF::Packed is data storage. We can read and write trivial (in C++ term [1]) data to this storage. The difference to
    the usual storage is that the alignment of this storage is always 1. We access the underlying data by using unalignedLoad/unalignedStore.
    This class offers alignment = 1 data structure instead of missing the following characteristics.

        1. Load / Store are non atomic even if the data size is within a pointer width. We should not use this for a member which can be accessed
           in a racy way. (e.g. fields accessed optimistically from the concurrent compilers).

        2. We cannot take reference / pointer to the underlying storage since they are unaligned.

        3. Access to this storage is unaligned access. The code is using memcpy, and the compiler will convert to an appropriate unaligned access
           in certain architectures (x86_64 / ARM64). It could be slow. So use it for non performance sensitive & memory sensitive places.

- WTF::PackedPtr

    WTF::PackedPtr is a specialization of WTF::Packed<T*>. And it is basically WTF::PackedAlignedPtr with alignment = 1. We further compact
    the pointer by leveraging the platform specific knowledge. In 64bit architectures, the effective width of pointers are less than 64 bit.
    In x86_64, it is 48 bits. And Darwin ARM64 is further smaller, 36 bits. This information allows us to compact the pointer to 6 bytes in
    x86_64 and 5 bytes in Darwin ARM64.

- WTF::PackedAlignedPtr

    WTF::PackedAlignedPtr is the WTF::PackedPtr with alignment information of the T. If we use this alignment information, we could reduce the
    size of packed pointer further in some cases. For example, since we guarantee that JSCells are 16 byte aligned, low 4 bits are empty. Leveraging
    this information in Darwin ARM64 platform allows us to make packed JSCell pointer 4 bytes (36 - 4 bits). We do not use passed alignment
    information if it is not profitable.

We also have PackedPtrTraits. This is new PtrTraits and use it for various data structures such as Bag<>.

[1]: https://en.cppreference.com/w/cpp/types/is_trivial

* WTF.xcodeproj/project.pbxproj:
* wtf/Bag.h:
(WTF::Bag::clear):
(WTF::Bag::iterator::operator++):
* wtf/CMakeLists.txt:
* wtf/DumbPtrTraits.h:
* wtf/DumbValueTraits.h:
* wtf/MathExtras.h:
(WTF::clzConstexpr):
(WTF::clz):
(WTF::ctzConstexpr):
(WTF::ctz):
(WTF::getLSBSetConstexpr):
(WTF::getMSBSetConstexpr):
* wtf/Packed.h: Added.
(WTF::Packed::Packed):
(WTF::Packed::get const):
(WTF::Packed::set):
(WTF::Packed::operator=):
(WTF::Packed::exchange):
(WTF::Packed::swap):
(WTF::alignof):
(WTF::PackedPtrTraits::exchange):
(WTF::PackedPtrTraits::swap):
(WTF::PackedPtrTraits::unwrap):
* wtf/Platform.h:
* wtf/SentinelLinkedList.h:
(WTF::BasicRawSentinelNode::BasicRawSentinelNode):
(WTF::BasicRawSentinelNode::prev):
(WTF::BasicRawSentinelNode::next):
(WTF::PtrTraits>::remove):
(WTF::PtrTraits>::prepend):
(WTF::PtrTraits>::append):
(WTF::RawNode>::SentinelLinkedList):
(WTF::RawNode>::remove):
(WTF::BasicRawSentinelNode<T>::remove): Deleted.
(WTF::BasicRawSentinelNode<T>::prepend): Deleted.
(WTF::BasicRawSentinelNode<T>::append): Deleted.
* wtf/StdLibExtras.h:
(WTF::roundUpToMultipleOfImpl):
(WTF::roundUpToMultipleOfImpl0): Deleted.
* wtf/UnalignedAccess.h:
(WTF::unalignedLoad):
(WTF::unalignedStore):

Tools:

* TestWebKitAPI/CMakeLists.txt:
* TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
* TestWebKitAPI/Tests/WTF/MathExtras.cpp:
(TestWebKitAPI::TEST):
* TestWebKitAPI/Tests/WTF/Packed.cpp: Added.
(TestWebKitAPI::TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245214 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[JSC] Compact generator code's bytecode size
ysuzuki@apple.com [Sun, 12 May 2019 19:43:16 +0000 (19:43 +0000)]
[JSC] Compact generator code's bytecode size
https://bugs.webkit.org/show_bug.cgi?id=197822

Reviewed by Michael Saboff.

op_put_to_scope's symbolTableOrScopeDepth is represented as int. This was OK for the old bytecode format since
VirtualRegister / scope depth can be represented by int anyway. But it is problematic now since only int8_t range
will be represented in narrow bytecode. When this field is used for symbol table constant index, it is always
larger than FirstConstantRegisterIndex. So it always exceeds the range of int8_t, and results in wide bytecode.
It makes all generator's op_put_to_scope wide bytecode.

In this patch, we introduce a new (logically) union type SymbolTableOrScopeDepth. It holds unsigned value, and we store the
SymbolTableConstantIndex - FirstConstantRegisterIndex in this unsigned value to make op_put_to_scope narrow bytecode.

* CMakeLists.txt:
* JavaScriptCore.xcodeproj/project.pbxproj:
* bytecode/BytecodeGeneratorification.cpp:
(JSC::BytecodeGeneratorification::run):
* bytecode/BytecodeList.rb:
* bytecode/CodeBlock.cpp:
(JSC::CodeBlock::finishCreation):
* bytecode/Fits.h:
* bytecompiler/BytecodeGenerator.cpp:
(JSC::BytecodeGenerator::BytecodeGenerator):
(JSC::BytecodeGenerator::emitProfileType):
(JSC::BytecodeGenerator::emitPutToScope):
(JSC::BytecodeGenerator::localScopeDepth const):
* bytecompiler/BytecodeGenerator.h:
* runtime/SymbolTableOrScopeDepth.h: Added.
(JSC::SymbolTableOrScopeDepth::symbolTable):
(JSC::SymbolTableOrScopeDepth::scopeDepth):
(JSC::SymbolTableOrScopeDepth::raw):
(JSC::SymbolTableOrScopeDepth::symbolTable const):
(JSC::SymbolTableOrScopeDepth::scopeDepth const):
(JSC::SymbolTableOrScopeDepth::raw const):
(JSC::SymbolTableOrScopeDepth::dump const):
(JSC::SymbolTableOrScopeDepth::SymbolTableOrScopeDepth):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245213 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoOverflow scroll that becomes non-scrollable should stop being composited
simon.fraser@apple.com [Sat, 11 May 2019 23:17:17 +0000 (23:17 +0000)]
Overflow scroll that becomes non-scrollable should stop being composited
https://bugs.webkit.org/show_bug.cgi?id=197817
<rdar://problem/50697290>

Reviewed by Antti Koivisto.

Source/WebCore:

Remove the iOS-specific #ifdef around code that triggers a compositing re-evaluation
when scrolling state changes.

Test: compositing/scrolling/async-overflow-scrolling/become-non-scrollable.html

* rendering/RenderLayer.cpp:
(WebCore::RenderLayer::updateScrollInfoAfterLayout):

LayoutTests:

Put in a compositing/scrolling/async-overflow-scrolling dir so we can enable it only
for platforms that have async overflow scrolling.

* TestExpectations:
* compositing/scrolling/async-overflow-scrolling/become-non-scrollable-expected.txt: Added.
* compositing/scrolling/async-overflow-scrolling/become-non-scrollable.html: Added.
* platform/ios-wk2/TestExpectations:
* platform/mac-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245212 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoCleaning up results for legacy-animation-engine tests
ap@apple.com [Sat, 11 May 2019 23:14:03 +0000 (23:14 +0000)]
Cleaning up results for legacy-animation-engine tests

Two tests were fixed in bug 188357.
The third one is still broken, and we have bug 184611 tracking it.

* platform/mac-wk1/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245211 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoRemoving flaky expectation for fast/repaint/canvas-object-fit.html
ap@apple.com [Sat, 11 May 2019 23:02:51 +0000 (23:02 +0000)]
Removing flaky expectation for fast/repaint/canvas-object-fit.html
https://bugs.webkit.org/show_bug.cgi?id=188910

* platform/mac-wk1/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245210 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoRemoving flaky expectation for fast/animation/request-animation-frame.html
ap@apple.com [Sat, 11 May 2019 22:29:21 +0000 (22:29 +0000)]
Removing flaky expectation for fast/animation/request-animation-frame.html
https://bugs.webkit.org/show_bug.cgi?id=190888

* platform/ios-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245209 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoLayer bounds are incorrect for sharing layers that paint with transforms
simon.fraser@apple.com [Sat, 11 May 2019 22:22:34 +0000 (22:22 +0000)]
Layer bounds are incorrect for sharing layers that paint with transforms
https://bugs.webkit.org/show_bug.cgi?id=197768
<rdar://problem/50695493>

Reviewed by Zalan Bujtas.

Source/WebCore:

We don't need to traverse shared layers if the backing-provider has overflow clip,
because we know they are containing-block descendants and therefore clipped.

Note tha the CSS "clip" property doesn't guarantee this, because the clip rect
can be larger than the element, so in that case we just traverse shared layers.

Tests: compositing/shared-backing/sharing-bounds-clip.html
       compositing/shared-backing/sharing-bounds-non-clipping-shared-layer.html
       compositing/shared-backing/sharing-bounds-transformed-sharing-layer.html
       compositing/shared-backing/sharing-bounds.html

* rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::updateCompositedBounds):

LayoutTests:

Tests for backing-shared layer bounds in various configurations.

* compositing/shared-backing/sharing-bounds-clip-expected.txt: Added.
* compositing/shared-backing/sharing-bounds-clip.html: Added.
* compositing/shared-backing/sharing-bounds-expected.txt: Added.
* compositing/shared-backing/sharing-bounds-non-clipping-shared-layer-expected.txt: Added.
* compositing/shared-backing/sharing-bounds-non-clipping-shared-layer.html: Added.
* compositing/shared-backing/sharing-bounds-transformed-sharing-layer-expected.txt: Added.
* compositing/shared-backing/sharing-bounds-transformed-sharing-layer.html: Added.
* compositing/shared-backing/sharing-bounds.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245208 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoTranslucent gradient rendering bug due to will-change transform
simon.fraser@apple.com [Sat, 11 May 2019 22:05:37 +0000 (22:05 +0000)]
Translucent gradient rendering bug due to will-change transform
https://bugs.webkit.org/show_bug.cgi?id=197654
<rdar://problem/50547664>

Reviewed by Dean Jackson.
Source/WebCore:

We failed to re-evaluate 'contentsOpaque' when a background changed, because this
happened in updateGeometry() and that doesn't run for background changes.

However, 'contentsOpaque' also requires knowing about geometry because we have to
turn it off when there's subpixel positioning, and updateConfiguration()
runs before updateGeometry().

So compute m_hasSubpixelRounding in updateGeometry() and set contentsOpaque in
updateAfterDescendants().

Test: compositing/contents-opaque/background-change-to-transparent.html

* rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::updateConfiguration):
(WebCore::RenderLayerBacking::updateGeometry):
(WebCore::RenderLayerBacking::updateAfterDescendants):
* rendering/RenderLayerBacking.h:

LayoutTests:

* compositing/contents-opaque/background-change-to-transparent-expected.txt: Added.
* compositing/contents-opaque/background-change-to-transparent.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245207 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoWhen the scroller hosting a shared layer becomes non-scrollable, content disappears
simon.fraser@apple.com [Sat, 11 May 2019 20:58:52 +0000 (20:58 +0000)]
When the scroller hosting a shared layer becomes non-scrollable, content disappears
https://bugs.webkit.org/show_bug.cgi?id=197766
<rdar://problem/50695808>

Reviewed by Zalan Bujtas.

Source/WebCore:

RenderLayerCompositor::requiresOwnBackingStore() should return true for a layer that shares
its backing store. We always made backing for overlap layers, so even  if the sharing layers
have no painted content, this should rarely be a backing store memory regression.

Test: compositing/shared-backing/overflow-scroll/sharing-layer-becomes-non-scrollable.html

* rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::requiresOwnBackingStore const):

LayoutTests:

* compositing/shared-backing/overflow-scroll/sharing-layer-becomes-non-scrollable-expected.html: Added.
* compositing/shared-backing/overflow-scroll/sharing-layer-becomes-non-scrollable.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245206 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoBacking-sharing layers with transforms render incorrectly
simon.fraser@apple.com [Sat, 11 May 2019 17:35:30 +0000 (17:35 +0000)]
Backing-sharing layers with transforms render incorrectly
https://bugs.webkit.org/show_bug.cgi?id=197692
<rdar://problem/50652127>

Reviewed by Antti Koivisto.
Source/WebCore:

Layers that paint into shared backing need to enter the RenderLayer painting code
in a way that paints the filters, transforms, opacity and blending.

RenderLayerBacking::paintIntoLayer() normally enters at paintLayerContents(), because
the effects are rendered via the GraphicsLayer, but shared layers will paint effects.
Note that if the backing-provider has effects, it will be the stacking context
for the shared layers, so it's correct that sharing layers are impacted by effects
on the backing-provider.

In addition, we have to ensure that we don't over-eagerly make layers shared.
Consider:

<div class="clipping">
    <div class="sharing">
        <div class="inner">
        </div>
    </div>
</div>

Here "clipping" is the provider layer, "sharing" paints into shared backing, but
we don't want to also mark "inner" as sharing, since "sharing" will just paint it.
This is akin to avoiding unnecessary compositing of z-order descendants when they can just
paint.

To do this we need to ensure that sharing layers are treated like compositing layers
in the overlap map, i.e. when a layer is sharing, we call overlapMap.pushCompositingContainer(),
and later overlapMap.popCompositingContainer().

Tests: compositing/shared-backing/nested-shared-layers-with-opacity.html
       compositing/shared-backing/shared-layer-has-blending.html
       compositing/shared-backing/shared-layer-has-filter.html
       compositing/shared-backing/shared-layer-has-opacity.html
       compositing/shared-backing/shared-layer-has-reflection.html
       compositing/shared-backing/shared-layer-has-transform.html
       compositing/shared-backing/shared-layer-isolates-blending.html
       compositing/shared-backing/shared-transformed-layer-bounds.html
       compositing/shared-backing/sharing-layer-becomes-non-scrollable.html
       compositing/shared-backing/sharing-layer-has-effect.html

* rendering/RenderLayer.cpp:
(WebCore::RenderLayer::paintLayer):
(WebCore::RenderLayer::paintLayerWithEffects):
* rendering/RenderLayer.h:
* rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::paintIntoLayer):
* rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::CompositingState::stateForPaintOrderChildren const):
(WebCore::backingProviderLayerCanIncludeLayer):
(WebCore::RenderLayerCompositor::computeCompositingRequirements):
(WebCore::RenderLayerCompositor::traverseUnchangedSubtree):

LayoutTests:

Ref tests for effects on sharing layers. The references make "clipping" be stacking context via z-index,
which eliminates sharing.

* compositing/shared-backing/nested-shared-layers-with-opacity-expected.html: Added.
* compositing/shared-backing/nested-shared-layers-with-opacity.html: Added.
* compositing/shared-backing/shared-layer-has-blending-expected.html: Added.
* compositing/shared-backing/shared-layer-has-blending.html: Added.
* compositing/shared-backing/shared-layer-has-filter-expected.html: Added.
* compositing/shared-backing/shared-layer-has-filter.html: Added.
* compositing/shared-backing/shared-layer-has-opacity-expected.html: Added.
* compositing/shared-backing/shared-layer-has-opacity.html: Added.
* compositing/shared-backing/shared-layer-has-reflection-expected.html: Added.
* compositing/shared-backing/shared-layer-has-reflection.html: Added.
* compositing/shared-backing/shared-layer-has-transform-expected.html: Added.
* compositing/shared-backing/shared-layer-has-transform.html: Added.
* compositing/shared-backing/shared-layer-isolates-blending-expected.html: Added.
* compositing/shared-backing/shared-layer-isolates-blending.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245205 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoUnreviewed, fix ProcessSwap.OpenerLinkAfterAPIControlledProcessSwappingOfOpener API...
cdumez@apple.com [Sat, 11 May 2019 02:15:21 +0000 (02:15 +0000)]
Unreviewed, fix ProcessSwap.OpenerLinkAfterAPIControlledProcessSwappingOfOpener API test

It became flaky after r245198.

* TestWebKitAPI/Tests/WebKitCocoa/ProcessSwapOnNavigation.mm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245204 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoCall to JSToWasmICCallee::createStructure passes in wrong prototype value
sbarati@apple.com [Sat, 11 May 2019 01:40:00 +0000 (01:40 +0000)]
Call to JSToWasmICCallee::createStructure passes in wrong prototype value
https://bugs.webkit.org/show_bug.cgi?id=197807
<rdar://problem/50530400>

Reviewed by Yusuke Suzuki.

JSTests:

* stress/js-to-wasm-callee-has-correct-prototype.js: Added.
(test.getInstance):
(test):

Source/JavaScriptCore:

We were passing the empty value instead of null. However, the empty
value means the Structure is poly proto. That's definitely not the case
here.

* runtime/JSGlobalObject.cpp:
(JSC::JSGlobalObject::init):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245203 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoBag's move operator= leaks memory
sbarati@apple.com [Sat, 11 May 2019 00:39:58 +0000 (00:39 +0000)]
Bag's move operator= leaks memory
https://bugs.webkit.org/show_bug.cgi?id=197757

Reviewed by Keith Miller.

It was unused. So I'm just removing it. We can implement it properly
if we ever need it.

* wtf/Bag.h:
(WTF::Bag::operator=): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245202 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[Test262] Unreviewed expectations update following r245188.
ross.kirsling@sony.com [Fri, 10 May 2019 23:42:53 +0000 (23:42 +0000)]
[Test262] Unreviewed expectations update following r245188.

* test262/config.yaml:
* test262/expectations.yaml:

* test262/test/intl402/DateTimeFormat/prototype/formatRange/date-is-infinity-throws.js:
* test262/test/intl402/DateTimeFormat/prototype/formatRange/date-is-nan-throws.js:
* test262/test/intl402/DateTimeFormat/prototype/formatRange/date-undefined-throws.js:
* test262/test/intl402/DateTimeFormat/prototype/formatRange/date-x-greater-than-y-throws.js:
* test262/test/intl402/DateTimeFormat/prototype/formatRange/this-is-not-object-throws.js:
* test262/test/intl402/DateTimeFormat/prototype/formatRangeToParts/date-is-infinity-throws.js:
* test262/test/intl402/DateTimeFormat/prototype/formatRangeToParts/date-is-nan-throws.js:
* test262/test/intl402/DateTimeFormat/prototype/formatRangeToParts/date-undefined-throws.js:
* test262/test/intl402/DateTimeFormat/prototype/formatRangeToParts/date-x-greater-than-y-throws.js:
* test262/test/intl402/DateTimeFormat/prototype/formatRangeToParts/this-is-not-object-throws.js:
These files have invalid YAML comments. Will also submit corrections back to Test262.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245201 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoA service worker instance should be terminated when its SWServer is destroyed
youenn@apple.com [Fri, 10 May 2019 23:33:38 +0000 (23:33 +0000)]
A service worker instance should be terminated when its SWServer is destroyed
https://bugs.webkit.org/show_bug.cgi?id=197801
<rdar://problem/50587270>

Reviewed by Chris Dumez.

Source/WebCore:

On session destruction, the SWServer is destroyed.
At that time, it should terminate all its running service workers.
Covered by updated API test.

* workers/service/server/SWServer.cpp:
(WebCore::SWServer::~SWServer):

Tools:

* TestWebKitAPI/Tests/WebKitCocoa/ServiceWorkerBasic.mm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245200 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[iOS] HTMLMediaElement sometimes doesn't send 'webkitplaybacktargetavailabilitychange...
eric.carlson@apple.com [Fri, 10 May 2019 23:29:56 +0000 (23:29 +0000)]
[iOS] HTMLMediaElement sometimes doesn't send 'webkitplaybacktargetavailabilitychanged' event
https://bugs.webkit.org/show_bug.cgi?id=197793
<rdar://problem/46429187>

Reviewed by Jer Noble.

* html/MediaElementSession.cpp:
(WebCore::MediaElementSession::MediaElementSession): Initialize m_hasPlaybackTargets.

* platform/audio/ios/MediaSessionManagerIOS.mm:
(WebCore::MediaSessionManageriOS::externalOutputDeviceAvailableDidChange): Log
target availability.
(-[WebMediaSessionHelper startMonitoringAirPlayRoutes]): Call the client
externalOutputDeviceAvailableDidChange method after the AVRouteDetector is available.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245199 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[PSON] Prevent flashing when the process-swap is forced by the client
cdumez@apple.com [Fri, 10 May 2019 23:25:28 +0000 (23:25 +0000)]
[PSON] Prevent flashing when the process-swap is forced by the client
https://bugs.webkit.org/show_bug.cgi?id=197799

Reviewed by Geoffrey Garen.

When the process-swap is forced by the client, we would not construct a SuspendedPageProxy for
the previous page, which would cause a white/black flash upon navigation on macOS. The reason
we did not construct a SuspendedPageProxy is that it would be unsafe to keep the page around
in this case because other windows might have an opener link to the page when the swap is forced
and we need those opener / openee links to get severed.

The new approach to maintain the Web facing behavior without flashing is to create a suspended
page proxy for the previous page when the process swap is forced by the client. We then close
the page as soon as we can do so without flashing (when pageEnteredAcceleratedCompositingMode()
has been called).

* UIProcess/SuspendedPageProxy.cpp:
(WebKit::SuspendedPageProxy::SuspendedPageProxy):
(WebKit::SuspendedPageProxy::close):
(WebKit::SuspendedPageProxy::pageEnteredAcceleratedCompositingMode):
(WebKit::SuspendedPageProxy::closeWithoutFlashing):
(WebKit::SuspendedPageProxy::didProcessRequestToSuspend):
* UIProcess/SuspendedPageProxy.h:
* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::suspendCurrentPageIfPossible):
(WebKit::WebPageProxy::commitProvisionalPage):
* UIProcess/WebPageProxy.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245198 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoDowngrade RELEASE_ASSERT TO RELEASE_LOG_FAULT for SQLite Class A files
ggaren@apple.com [Fri, 10 May 2019 21:36:45 +0000 (21:36 +0000)]
Downgrade RELEASE_ASSERT TO RELEASE_LOG_FAULT for SQLite Class A files
https://bugs.webkit.org/show_bug.cgi?id=197760

Reviewed by Jer Noble.

We have all the data we need, and this crash is happening more than
expected.

* platform/sql/SQLiteDatabase.cpp:
(WebCore::SQLiteDatabase::open):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245197 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoStreamline test-and-clear operation for ContextMenu
bfulgham@apple.com [Fri, 10 May 2019 21:28:06 +0000 (21:28 +0000)]
Streamline test-and-clear operation for ContextMenu
https://bugs.webkit.org/show_bug.cgi?id=197795
<rdar://problem/50473746>

Reviewed by Wenson Hsieh.

Rather than performing a check when entering the function, doing work,
then clearing the member variable, perform the check and clear in
a single operation. Only perform the selection if the menu exists.

* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::didSelectItemFromActiveContextMenu):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245196 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoReturning incorrect marked text rects
megan_gardner@apple.com [Fri, 10 May 2019 20:52:32 +0000 (20:52 +0000)]
Returning incorrect marked text rects
https://bugs.webkit.org/show_bug.cgi?id=197758
<rdar://problem/46548586>

Reviewed by Tim Horton.

We are not returning marked text rects in the correct view coordinate space.
This is the same translation we are applying to the caret rects.

* WebProcess/WebPage/ios/WebPageIOS.mm:
(WebKit::WebPage::platformEditorState const):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245195 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[JSC] String substring operation should return ropes consistently
ysuzuki@apple.com [Fri, 10 May 2019 20:49:35 +0000 (20:49 +0000)]
[JSC] String substring operation should return ropes consistently
https://bugs.webkit.org/show_bug.cgi?id=197765
<rdar://problem/37689944>

Reviewed by Michael Saboff.

Currently we have different policies per string substring operation function.

    1. String#slice returns the resolved non-rope string
    2. String#substring returns rope string
    3. String#substr returns rope string in runtime function, non-rope string in DFG and FTL

Due to (3), we see large memory use in the tested web page[1]. Non rope substring have a problem.
First of all, that returned string seems not used immediately. It is possible that the resulted
string is used as a part of the other ropes (like, xxx.substring(...) + "Hello"). To avoid the
eager materialization of the string, we are using StringImpl::createSubstringSharingImpl for the
resulted non rope string. StringImpl::createSubstringSharingImpl is StringImpl's substring feature: the
substring is pointing the owner StringImpl. While this has memory saving benefit, it can retain owner
StringImpl so long, and it could keep very large owner StringImpl alive.

The problem we are attempting to solve with StringImpl::createSubstringSharingImpl can be solved by
the rope string simply. Rope string can share the underlying string. And good feature of the rope
string is that, when resolving rope string, the rope string always create a new StringImpl instead of
using StringImpl::createSubstringSharingImpl. So we allow the owner StringImpl to be destroyed. And this
resolving only happens when we actually want to use the content of the rope string. In addition, we recently
shrunk the sizeof(JSRopeString) from 48 to 32, so JSRopeString is cheap.

In this patch, we change (2) and (3) to (1), using rope String as a result of substring operations.

RAMification and JetStream2 are neutral. The web page[1] shows large memory footprint improvement from 776MB to 681MB.

[1]: https://beta.observablehq.com/@ldgardner/assignment-4-visualizations-and-multiple-views

* dfg/DFGOperations.cpp:
* runtime/StringPrototype.cpp:
(JSC::stringProtoFuncSlice):
* runtime/StringPrototypeInlines.h:
(JSC::stringSlice):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245194 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agotestb3 failing with crash in JSC::B3::BasicBlock::appendNonTerminal
rmorisset@apple.com [Fri, 10 May 2019 20:37:07 +0000 (20:37 +0000)]
testb3 failing with crash in JSC::B3::BasicBlock::appendNonTerminal
https://bugs.webkit.org/show_bug.cgi?id=197756
<rdar://problem/50641659>

Reviewed by Saam Barati.

When I added https://bugs.webkit.org/show_bug.cgi?id=197265 I assumed that which block is the root does not change in the middle of strength reduction.
But specializeSelect can use splitForward, which uses a new block for the first half of the given block.
So if the block being split is the root block I must update m_root and erase the m_valueInConstant cache.
Erasing the cache cannot cause wrong results: at most it can make us miss some optimization opportunities in this iteration of the fixpoint.

* b3/B3ReduceStrength.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245192 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[iOS] baidu.com: Synthetic bold renders too far apart, appears doubled.
zalan@apple.com [Fri, 10 May 2019 20:30:05 +0000 (20:30 +0000)]
[iOS] baidu.com: Synthetic bold renders too far apart, appears doubled.
https://bugs.webkit.org/show_bug.cgi?id=197781
<rdar://problem/48027412>

Reviewed by Simon Fraser.

Source/WebCore:

Synthetic bold is essentially two regular glyphs painted with an offset. While on macOS this offset is always 1px (CSS), on iOS larger font produces higher offset value. At paint time, this offset value (in CSS px unit) get converted
to a device pixel value taking context scale into account. This conversion ensures that the gap between the 2 regular glyphs won't get wider (in device pixels) as the user pinch zooms in.
This works as long as the scale on the context is >= 1. This patch ensures that a scaled down context won't blow up this gap.

Test: fast/text/large-synthetic-bold-with-scale-transform.html

* platform/graphics/cocoa/FontCascadeCocoa.mm:
(WebCore::FontCascade::drawGlyphs):

LayoutTests:

* fast/text/large-synthetic-bold-with-scale-transform-expected.html: Added.
* fast/text/large-synthetic-bold-with-scale-transform.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245191 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoGracefully handle inaccessible font face data
bfulgham@apple.com [Fri, 10 May 2019 20:10:59 +0000 (20:10 +0000)]
Gracefully handle inaccessible font face data
https://bugs.webkit.org/show_bug.cgi?id=197762
<rdar://problem/50433861>

Reviewed by Per Arne Vollan.

Make sure CSS Font Face handling gracefully recovers from
missing font data.

Test: fast/text/missing-font-crash.html

* css/CSSFontFace.cpp:
(WebCore::CSSFontFace::fontLoadEventOccurred):
(WebCore::CSSFontFace::timeoutFired):
(WebCore::CSSFontFace::fontLoaded):
(WebCore::CSSFontFace::font):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245190 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoLet WebKit clients supply an application name for user agent in modern compatibility...
wenson_hsieh@apple.com [Fri, 10 May 2019 19:58:37 +0000 (19:58 +0000)]
Let WebKit clients supply an application name for user agent in modern compatibility mode
https://bugs.webkit.org/show_bug.cgi?id=197782
<rdar://problem/50646747>

Reviewed by Beth Dakin.

Add plumbing for a new private property on WKWebpagePreferences.

* UIProcess/API/APIWebsitePolicies.cpp:
(API::WebsitePolicies::copy const):
* UIProcess/API/APIWebsitePolicies.h:
* UIProcess/API/Cocoa/WKWebpagePreferences.mm:
(-[WKWebpagePreferences _applicationNameForUserAgentWithModernCompatibility]):
(-[WKWebpagePreferences _setApplicationNameForUserAgentWithModernCompatibility:]):
* UIProcess/API/Cocoa/WKWebpagePreferencesPrivate.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245189 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoUpdate test262 tests.
keith_miller@apple.com [Fri, 10 May 2019 19:45:37 +0000 (19:45 +0000)]
Update test262 tests.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245188 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoThe active tab sometimes app naps even though it should not
cdumez@apple.com [Fri, 10 May 2019 19:41:36 +0000 (19:41 +0000)]
The active tab sometimes app naps even though it should not
https://bugs.webkit.org/show_bug.cgi?id=197791
<rdar://problem/48460054>

Reviewed by Geoffrey Garen.

The WebPage constructor was initializing m_activityState but not calling updateThrottleState().
As a result, we would not take a UserActivity even when warranted by the initial activity state.

* WebProcess/WebPage/WebPage.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245187 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[WinCairo] storage/indexeddb tests are timing out
Hironori.Fujii@sony.com [Fri, 10 May 2019 19:01:54 +0000 (19:01 +0000)]
[WinCairo] storage/indexeddb tests are timing out
https://bugs.webkit.org/show_bug.cgi?id=196289

Reviewed by Alex Christensen.

Source/WebKit:

storage/indexeddb tests were timing out for WinCairo port because
WebKit::NetworkCache classes were not implemented yet for Windows.

Implement WebKit::NetworkCache classes by using WTF::FileSystem
functions.

* NetworkProcess/cache/CacheStorageEngine.cpp:
(WebKit::CacheStorage::Engine::readFile): Use
IOChannel::isOpened() to check the channel is opened instead of
checking the file descriptor.
* NetworkProcess/cache/NetworkCacheBlobStorage.cpp:
(WebKit::NetworkCache::BlobStorage::add):
(WebKit::NetworkCache::BlobStorage::remove):
* NetworkProcess/cache/NetworkCacheData.cpp:
(WebKit::NetworkCache::Data::mapToFile const):
(WebKit::NetworkCache::mapFile):
(WebKit::NetworkCache::adoptAndMapFile):
(WebKit::NetworkCache::makeSalt):
(WebKit::NetworkCache::readOrMakeSalt):
* NetworkProcess/cache/NetworkCacheData.h:
(WebKit::NetworkCache::Data::isEmpty const):
(WebKit::NetworkCache::Data::size const):
* NetworkProcess/cache/NetworkCacheDataCurl.cpp:
(WebKit::NetworkCache::Data::Data):
(WebKit::NetworkCache::Data::empty):
(WebKit::NetworkCache::Data::data const):
(WebKit::NetworkCache::Data::isNull const):
(WebKit::NetworkCache::Data::apply const):
(WebKit::NetworkCache::Data::subrange const):
(WebKit::NetworkCache::concatenate):
(WebKit::NetworkCache::Data::adoptMap): Deleted.
* NetworkProcess/cache/NetworkCacheFileSystem.cpp:
(WebKit::NetworkCache::traverseDirectory):
(WebKit::NetworkCache::fileTimes):
(WebKit::NetworkCache::updateFileModificationTimeIfNeeded):
(WebKit::NetworkCache::isSafeToUseMemoryMapForPath):
* NetworkProcess/cache/NetworkCacheIOChannel.h:
(WebKit::NetworkCache::IOChannel::isOpened const):
(WebKit::NetworkCache::IOChannel::fileDescriptor const): Deleted.
* NetworkProcess/cache/NetworkCacheIOChannelCurl.cpp:
(WebKit::NetworkCache::IOChannel::IOChannel):
(WebKit::NetworkCache::IOChannel::~IOChannel):
(WebKit::NetworkCache::runTaskInQueue):
(WebKit::NetworkCache::IOChannel::read):
(WebKit::NetworkCache::IOChannel::write):

Source/WTF:

* wtf/FileSystem.h: Added hardLink.
* wtf/glib/FileSystemGlib.cpp:
(WTF::FileSystemImpl::hardLink):
(WTF::FileSystemImpl::hardLinkOrCopyFile):
* wtf/posix/FileSystemPOSIX.cpp:
(WTF::FileSystemImpl::hardLink):
(WTF::FileSystemImpl::hardLinkOrCopyFile):
* wtf/win/FileSystemWin.cpp:
(WTF::FileSystemImpl::hardLink):
(WTF::FileSystemImpl::hardLinkOrCopyFile):
Added hardLink. Let hardLinkOrCopyFile use the hardLink.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245186 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoDo not wait until requestPermission() is called to fire deviceorientation events...
cdumez@apple.com [Fri, 10 May 2019 19:01:12 +0000 (19:01 +0000)]
Do not wait until requestPermission() is called to fire deviceorientation events if permission was already granted
https://bugs.webkit.org/show_bug.cgi?id=197750

Reviewed by Geoffrey Garen.

Source/WebKit:

The UIProcess remembers previous device orientation permission decisions per origin for the duration of the browsing
session. However, the WebContent process was not aware of previous decisions and would therefore not fire any
deviceorientation / devicemotion events until the JS has called requestPermission(). This patches addresses this
problem by having the UIProcess communicate any previous permission decision for the origin via WebSitePolicies.

* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::receivedNavigationPolicyDecision):
* UIProcess/WebsiteData/WebDeviceOrientationAndMotionAccessController.cpp:
(WebKit::WebDeviceOrientationAndMotionAccessController::shouldAllowAccess):
(WebKit::WebDeviceOrientationAndMotionAccessController::cachedDeviceOrientationPermission const):
(WebKit::WebDeviceOrientationAndMotionAccessController::deviceOrientationPermission const): Deleted.
* UIProcess/WebsiteData/WebDeviceOrientationAndMotionAccessController.h:

Tools:

Add API test coverage.

* TestWebKitAPI/Tests/WebKitCocoa/DeviceOrientation.mm:
(TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245185 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoRemove legacy test262 import script
keith_miller@apple.com [Fri, 10 May 2019 18:56:08 +0000 (18:56 +0000)]
Remove legacy test262 import script
https://bugs.webkit.org/show_bug.cgi?id=197788

Reviewed by Yusuke Suzuki.

import-test262-tests doesn't work anymore. We should be using test262-import instead.

* Scripts/import-test262-tests: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245184 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoAdd WKWebViewConfiguration._canShowWhileLocked SPI
cdumez@apple.com [Fri, 10 May 2019 18:52:05 +0000 (18:52 +0000)]
Add WKWebViewConfiguration._canShowWhileLocked SPI
https://bugs.webkit.org/show_bug.cgi?id=197777
<rdar://problem/50065100>

Reviewed by Tim Horton.

Source/WebCore/PAL:

* pal/spi/cocoa/QuartzCoreSPI.h:

Source/WebKit:

* Platform/mac/LayerHostingContext.h:
* Platform/mac/LayerHostingContext.mm:
(WebKit::LayerHostingContext::createForExternalHostingProcess):
* Shared/WebPageCreationParameters.cpp:
(WebKit::WebPageCreationParameters::encode const):
(WebKit::WebPageCreationParameters::decode):
* Shared/WebPageCreationParameters.h:
* UIProcess/API/APIPageConfiguration.cpp:
(API::PageConfiguration::copy const):
* UIProcess/API/APIPageConfiguration.h:
(API::PageConfiguration::canShowWhileLocked const):
(API::PageConfiguration::setCanShowWhileLocked):
* UIProcess/API/Cocoa/WKWebViewConfiguration.mm:
(-[WKWebViewConfiguration _setCanShowWhileLocked:]):
(-[WKWebViewConfiguration _canShowWhileLocked]):
* UIProcess/API/Cocoa/WKWebViewConfigurationPrivate.h:
* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::creationParameters):
* WebProcess/WebPage/RemoteLayerTree/PlatformCALayerRemoteCustom.mm:
(WebKit::PlatformCALayerRemoteCustom::PlatformCALayerRemoteCustom):
* WebProcess/WebPage/RemoteLayerTree/RemoteLayerTreeContext.h:
* WebProcess/WebPage/RemoteLayerTree/RemoteLayerTreeContext.mm:
(WebKit::RemoteLayerTreeContext::canShowWhileLocked const):
* WebProcess/WebPage/WebPage.cpp:
(WebKit::m_canShowWhileLocked):
(WebKit::m_shrinkToFitContentTimer): Deleted.
* WebProcess/WebPage/WebPage.h:
(WebKit::WebPage::canShowWhileLocked const):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245183 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[WPE] Forward libepoxy cflags
cturner@igalia.com [Fri, 10 May 2019 18:46:55 +0000 (18:46 +0000)]
[WPE] Forward libepoxy cflags
https://bugs.webkit.org/show_bug.cgi?id=197784

Reviewed by Žan Doberšek.

* wpe/backends/CMakeLists.txt: This is required to keep
-DMESA_EGL_NO_X11_HEADERS in the build, if we don't do this, EGL ends
up trying to include X11 headers even when they're not present in the
environment.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245182 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoASSERT(isSelfPaintingLayer() || hasSelfPaintingLayerDescendant()) on nytimes.com...
simon.fraser@apple.com [Fri, 10 May 2019 18:40:58 +0000 (18:40 +0000)]
ASSERT(isSelfPaintingLayer() || hasSelfPaintingLayerDescendant()) on nytimes.com after r245170
https://bugs.webkit.org/show_bug.cgi?id=197776

Reviewed by Zalan Bujtas.

Source/WebCore:

Only try to paint into shared backing for layers that are able to be composited. This
avoids trying to do sharing for non-self-painting layers, which doesn't make sense.

Test: compositing/shared-backing/overflow-scroll/non-self-painting-layer-should-not-share.html

* rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::computeCompositingRequirements):

LayoutTests:

* compositing/shared-backing/overflow-scroll/non-self-painting-layer-should-not-share-expected.txt: Added.
* compositing/shared-backing/overflow-scroll/non-self-painting-layer-should-not-share.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245181 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoFix a bunch of compiler warnings
mcatanzaro@igalia.com [Fri, 10 May 2019 18:03:56 +0000 (18:03 +0000)]
Fix a bunch of compiler warnings
https://bugs.webkit.org/show_bug.cgi?id=197785

Reviewed by Don Olmstead.

Source/ThirdParty/ANGLE:

Add compiler flags to suppress various warnings triggered by the new version of ANGLE.

* CMakeLists.txt:

Source/WebCore:

* CMakeLists.txt: WebCoreTestSupport should accept extra SYSTEM includes.
* PlatformGTK.cmake: GTK includes should be added to SYSTEM headers to avoid warnings.
* inspector/InspectorCanvas.cpp: Add preprocessor guards to fix unused function warning.
* rendering/RenderLayer.cpp: Fix unused variable warning.
(WebCore::RenderLayer::updateClipRects):

Source/WebKit:

Guard Cocoa-specific format modifier to avoid -Wformat warning.

* NetworkProcess/AdClickAttributionManager.cpp:
(WebKit::AdClickAttributionManager::fireConversionRequest):

Tools:

Add preprocessor guards to avoid -Wunused-function warnings.

* TestWebKitAPI/Tests/WebCore/FloatRect.cpp:
* TestWebKitAPI/Tests/WebCore/IntRect.cpp:
* TestWebKitAPI/Tests/WebCore/TransformationMatrix.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245179 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoAX: Crash at WebKit: WebKit::WebSpeechSynthesisClient::speak
cfleizach@apple.com [Fri, 10 May 2019 17:19:24 +0000 (17:19 +0000)]
AX: Crash at WebKit: WebKit::WebSpeechSynthesisClient::speak
https://bugs.webkit.org/show_bug.cgi?id=197761
<rdar://problem/50237614>

Reviewed by Per Arne Vollan.

Source/WebKit:

It's valid for the incoming voice to be nil, so we need to protect against that.

Tests: fast/speechsynthesis/speech-synthesis-real-client-version.html

* WebProcess/WebCoreSupport/WebSpeechSynthesisClient.cpp:
(WebKit::WebSpeechSynthesisClient::speak):

LayoutTests:

* fast/speechsynthesis/speech-synthesis-real-client-version-expected.txt: Added.
* fast/speechsynthesis/speech-synthesis-real-client-version.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245178 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[WTF] Remove "private:" from Noncopyable and Nonmovable macros
ysuzuki@apple.com [Fri, 10 May 2019 16:56:31 +0000 (16:56 +0000)]
[WTF] Remove "private:" from Noncopyable and Nonmovable macros
https://bugs.webkit.org/show_bug.cgi?id=197767

Reviewed by Saam Barati.

We no longer need "private:".

* wtf/Noncopyable.h:
* wtf/Nonmovable.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245177 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[WPE][GTK] Add webkit_frame_get_id() API
mcatanzaro@igalia.com [Fri, 10 May 2019 16:46:39 +0000 (16:46 +0000)]
[WPE][GTK] Add webkit_frame_get_id() API
https://bugs.webkit.org/show_bug.cgi?id=197270

Reviewed by Carlos Garcia Campos.

Source/WebKit:

It's as simple as can be: just an API to return the ID of a frame.

* UIProcess/API/gtk/docs/webkit2gtk-4.0-sections.txt:
* WebProcess/InjectedBundle/API/glib/WebKitFrame.cpp:
(webkit_frame_get_id):
* WebProcess/InjectedBundle/API/gtk/WebKitFrame.h:
* WebProcess/InjectedBundle/API/wpe/WebKitFrame.h:
* WebProcess/InjectedBundle/API/wpe/docs/wpe-webextensions-1.0-sections.txt:

Tools:

This was hard, but I added a test to verify that two different WebKitFrames of the same
WebKitWebPage return two different frame IDs.

* TestWebKitAPI/Tests/WebKitGLib/FrameTest.cpp:
(WebKitFrameTest::willSubmitFormCallback):
(WebKitFrameTest::testSubframe):
(registerTests):
* TestWebKitAPI/Tests/WebKitGLib/TestFrame.cpp:
(testWebKitFrameSubframe):
(beforeAll):
* TestWebKitAPI/Tests/WebKitGLib/WebExtensionTest.cpp:
(willSubmitFormCallback):
* TestWebKitAPI/Tests/WebKitGLib/resources/form-in-frame.html: Added.
* TestWebKitAPI/Tests/WebKitGLib/resources/webkitglib-tests.gresource.xml:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245176 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoEvent region generation needs to know about backing-sharing
antti@apple.com [Fri, 10 May 2019 16:20:12 +0000 (16:20 +0000)]
Event region generation needs to know about backing-sharing
https://bugs.webkit.org/show_bug.cgi?id=197694
<rdar://problem/50584991>

Reviewed by Simon Fraser.

Source/WebCore:

Test: pointerevents/ios/touch-action-region-backing-sharing.html

* rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::updateEventRegion):

Gather event region from backing sharing layers too.

LayoutTests:

* platform/ios-wk2/TestExpectations:
* pointerevents/ios/touch-action-region-backing-sharing-expected.txt: Added.
* pointerevents/ios/touch-action-region-backing-sharing.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245175 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[WPE][GTK] Add WebKitWebPage::did-associate-form-controls-for-frame and deprecate...
mcatanzaro@igalia.com [Fri, 10 May 2019 16:05:34 +0000 (16:05 +0000)]
[WPE][GTK] Add WebKitWebPage::did-associate-form-controls-for-frame and deprecate original did-associate-form-controls
https://bugs.webkit.org/show_bug.cgi?id=197271

Reviewed by Youenn Fablet.

Source/WebCore:

* dom/Document.cpp:
(WebCore::Document::didAssociateFormControlsTimerFired):
* loader/EmptyClients.h:
* page/ChromeClient.h:

Source/WebKit:

* WebProcess/InjectedBundle/API/c/WKBundlePageFormClient.h:
* WebProcess/InjectedBundle/API/glib/WebKitWebPage.cpp:
(webkit_web_page_class_init):
* WebProcess/InjectedBundle/API/mac/WKWebProcessPlugInBrowserContextController.mm:
(-[WKWebProcessPlugInBrowserContextController _setFormDelegate:]):
* WebProcess/InjectedBundle/APIInjectedBundleFormClient.h:
(API::InjectedBundle::FormClient::didAssociateFormControls):
* WebProcess/InjectedBundle/InjectedBundlePageFormClient.cpp:
(WebKit::InjectedBundlePageFormClient::didAssociateFormControls):
* WebProcess/InjectedBundle/InjectedBundlePageFormClient.h:
* WebProcess/WebCoreSupport/WebChromeClient.cpp:
(WebKit::WebChromeClient::didAssociateFormControls):
* WebProcess/WebCoreSupport/WebChromeClient.h:

Tools:

* TestWebKitAPI/Tests/WebKitGLib/WebExtensionTest.cpp:
(formControlsAssociatedForFrameCallback):
(pageCreatedCallback):
(formControlsAssociatedCallback): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245174 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[iOS Debug] Layout Test webgl/2.0.0/conformance2/attribs/gl-vertexattribipointer...
ryanhaddad@apple.com [Fri, 10 May 2019 16:03:42 +0000 (16:03 +0000)]
[iOS Debug] Layout Test webgl/2.0.0/conformance2/attribs/gl-vertexattribipointer.html is a flaky timeout
https://bugs.webkit.org/show_bug.cgi?id=197778

Unreviewed test gardening.

* platform/ios/TestExpectations: Mark test as slow.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245173 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[Flatpak] Use the safe PyYAML loader if available
aperez@igalia.com [Fri, 10 May 2019 14:42:27 +0000 (14:42 +0000)]
[Flatpak] Use the safe PyYAML loader if available
https://bugs.webkit.org/show_bug.cgi?id=197771

Reviewed by Philippe Normand.

* flatpak/flatpakutils.py:
(load_manifest): Use yaml.safe_load() if available, with yaml.load()
still used as fallback when the former is not provided by the module.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245172 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoLimit number of prefetches of a given page
commit-queue@webkit.org [Fri, 10 May 2019 06:44:13 +0000 (06:44 +0000)]
Limit number of prefetches of a given page
https://bugs.webkit.org/show_bug.cgi?id=197736

Patch by Rob Buis <rbuis@igalia.com> on 2019-05-09
Reviewed by Darin Adler.

Limit prefetches for same url to 1.

* NetworkProcess/cache/PrefetchCache.cpp:
(WebKit::PrefetchCache::store):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245171 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoImplement backing-sharing in compositing layers, allowing overlap layers to paint...
simon.fraser@apple.com [Fri, 10 May 2019 05:18:02 +0000 (05:18 +0000)]
Implement backing-sharing in compositing layers, allowing overlap layers to paint into the backing store of another layer
https://bugs.webkit.org/show_bug.cgi?id=197561
<rdar://problem/50445998>

Reviewed by Antti Koivisto.
Source/WebCore:

This change introduces the concept of layers that share backing store for compositing. A layer
which is sharing its backing store first paints itself, and then some set of layers which come
later in paint order in the same stacking context. This reduces the composited layer count in
some overflow scrolling scenarios, thereby also simplifying the scrolling tree.

A backing-shared layer stores a vector of "sharing" RenderLayer* in its RenderLayerBacking. At
paint time, the owning layer is painted, then the sharing layers, setting the owning layer as the
painting root so that positioning and clipping just work.

Sharing layer relationships are constructed in RenderLayerCompositor::computeCompositingRequirements().
We track the last layer which was composited in paint order as a shared candidate. If a later layer
would composite for overlap (and no other reasons), then we allow it to share with the candidate
if the candidate is in its ancestor containing block chain. Sharing is currently limited to layers
in the same stacking context.

isComposited() returns false for sharing layers, but they are like composited layers in that
they behave as painting boundaries, so RenderLayer::paintLayer() needs to stop at them,
and repaints in shared layers have to be directed to their shared layer, hence
changes to RenderLayer::clippingRootForPainting() and RenderLayer::enclosingCompositingLayerForRepaint().

The clipping boundary logic in RenderLayer::backgroundClipRect() needed to be generalized so that
all calls to RenderLayer::parentClipRects() check for crossing painting boundaries and use
TemporaryClipRects in that case.

Tests: compositing/shared-backing/overflow-scroll/absolute-in-stacking-relative-in-scroller.html
       compositing/shared-backing/overflow-scroll/composited-absolute-in-absolute-in-relative-in-scroller.html
       compositing/shared-backing/overflow-scroll/nested-absolute-with-clipping-in-stacking-overflow.html
       compositing/shared-backing/overflow-scroll/previous-sibling-prevents-inclusiveness.html
       compositing/shared-backing/overflow-scroll/relative-in-clipping-in-scroller-in-clipping.html
       compositing/shared-backing/overflow-scroll/relative-in-clipping-in-scroller-in-relative-clipping.html
       compositing/shared-backing/overflow-scroll/relative-in-div-in-overflow-scroll.html
       compositing/shared-backing/overflow-scroll/scrolled-contents-has-painted-content.html
       compositing/shared-backing/overflow-scroll/scrolled-contents-unconstrained-clip.html
       compositing/shared-backing/overflow-scroll/shared-layer-clipping.html
       compositing/shared-backing/overflow-scroll/shared-layer-composited-bounds.html
       compositing/shared-backing/overflow-scroll/shared-layer-nested-relative-stacking.html
       compositing/shared-backing/overflow-scroll/shared-layer-repaint.html
       compositing/shared-backing/partial-compositing-update.html
       compositing/shared-backing/partial-compositing-update2.html
       compositing/shared-backing/remove-sharing-layer.html
       compositing/shared-backing/sharing-cached-clip-rects.html

* rendering/RenderLayer.cpp:
(WebCore::RenderLayer::~RenderLayer):
(WebCore::RenderLayer::ancestorLayerIsInContainingBlockChain const):
(WebCore::RenderLayer::setBackingProviderLayer):
(WebCore::RenderLayer::disconnectFromBackingProviderLayer):
(WebCore::RenderLayer::enclosingCompositingLayerForRepaint const):
(WebCore::RenderLayer::clippingRootForPainting const):
(WebCore::RenderLayer::clipToRect):
(WebCore::RenderLayer::paintLayer):
(WebCore::RenderLayer::updateClipRects):
(WebCore::RenderLayer::clipCrossesPaintingBoundary const):
(WebCore::RenderLayer::calculateClipRects const):
(WebCore::outputPaintOrderTreeLegend):
(WebCore::outputPaintOrderTreeRecursive):
(WebCore::inContainingBlockChain): Deleted.
* rendering/RenderLayer.h:
* rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::willBeDestroyed):
(WebCore::clearBackingSharingLayerProviders):
(WebCore::RenderLayerBacking::setBackingSharingLayers):
(WebCore::RenderLayerBacking::removeBackingSharingLayer):
(WebCore::RenderLayerBacking::clearBackingSharingLayers):
(WebCore::RenderLayerBacking::updateCompositedBounds):
(WebCore::RenderLayerBacking::updateDrawsContent):
(WebCore::RenderLayerBacking::isSimpleContainerCompositingLayer const):
(WebCore::RenderLayerBacking::paintIntoLayer):
(WebCore::RenderLayerBacking::paintContents):
* rendering/RenderLayerBacking.h:
* rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::CompositingState::stateForPaintOrderChildren const):
(WebCore::RenderLayerCompositor::CompositingState::propagateStateFromChildren):
(WebCore::RenderLayerCompositor::CompositingState::propagateStateFromChildrenForUnchangedSubtree):
(WebCore::RenderLayerCompositor::BackingSharingState::resetBackingProviderCandidate):
(WebCore::RenderLayerCompositor::updateCompositingLayers):
(WebCore::backingProviderLayerCanIncludeLayer):
(WebCore::RenderLayerCompositor::computeCompositingRequirements):
(WebCore::RenderLayerCompositor::traverseUnchangedSubtree):
(WebCore::RenderLayerCompositor::updateBacking):
(WebCore::RenderLayerCompositor::layerWillBeRemoved):
(WebCore::RenderLayerCompositor::requiresCompositingForIndirectReason const):
* rendering/RenderLayerCompositor.h:
* rendering/RenderTreeAsText.cpp:

LayoutTests:

New tests for backing sharing, and new baselines of tests whose behavior is changed.

Scrolling tree tests that would be invalidate by sharing are changed to defeat sharing by adding
compositing layers early in stacking order.

* TestExpectations:
* compositing/geometry/limit-layer-bounds-clipping-ancestor-expected.txt:
* compositing/layer-creation/overflow-scroll-overlap-expected.txt:
* compositing/layer-creation/overflow-scroll-overlap.html:
* compositing/overflow/scrolling-content-clip-to-viewport.html:
* compositing/rtl/rtl-scrolling-with-transformed-descendants-expected.txt:
* compositing/shared-backing/overflow-scroll/absolute-in-stacking-relative-in-scroller-expected.txt: Copied from LayoutTests/platform/ios/compositing/overflow/scrolling-content-clip-to-viewport-expected.txt.
* compositing/shared-backing/overflow-scroll/absolute-in-stacking-relative-in-scroller.html: Added.
* compositing/shared-backing/overflow-scroll/composited-absolute-in-absolute-in-relative-in-scroller-expected.txt: Added.
* compositing/shared-backing/overflow-scroll/composited-absolute-in-absolute-in-relative-in-scroller.html: Added.
* compositing/shared-backing/overflow-scroll/nested-absolute-with-clipping-in-stacking-overflow-expected.txt: Added.
* compositing/shared-backing/overflow-scroll/nested-absolute-with-clipping-in-stacking-overflow.html: Copied from LayoutTests/scrollingcoordinator/scrolling-tree/nested-absolute-in-sc-overflow.html.
* compositing/shared-backing/overflow-scroll/previous-sibling-prevents-inclusiveness-expected.txt: Added.
* compositing/shared-backing/overflow-scroll/previous-sibling-prevents-inclusiveness.html: Added.
* compositing/shared-backing/overflow-scroll/relative-in-clipping-in-scroller-in-clipping-expected.txt: Added.
* compositing/shared-backing/overflow-scroll/relative-in-clipping-in-scroller-in-clipping.html: Added.
* compositing/shared-backing/overflow-scroll/relative-in-clipping-in-scroller-in-relative-clipping-expected.txt: Added.
* compositing/shared-backing/overflow-scroll/relative-in-clipping-in-scroller-in-relative-clipping.html: Added.
* compositing/shared-backing/overflow-scroll/relative-in-div-in-overflow-scroll-expected.txt: Copied from LayoutTests/platform/ios/compositing/overflow/scrolling-content-clip-to-viewport-expected.txt.
* compositing/shared-backing/overflow-scroll/relative-in-div-in-overflow-scroll.html: Added.
* compositing/shared-backing/overflow-scroll/scrolled-contents-has-painted-content-expected.txt: Copied from LayoutTests/platform/ios/compositing/overflow/scrolling-content-clip-to-viewport-expected.txt.
* compositing/shared-backing/overflow-scroll/scrolled-contents-has-painted-content.html: Added.
* compositing/shared-backing/overflow-scroll/scrolled-contents-unconstrained-clip-expected.html: Added.
* compositing/shared-backing/overflow-scroll/scrolled-contents-unconstrained-clip.html: Added.
* compositing/shared-backing/overflow-scroll/shared-layer-clipping-expected.html: Added.
* compositing/shared-backing/overflow-scroll/shared-layer-clipping.html: Added.
* compositing/shared-backing/overflow-scroll/shared-layer-composited-bounds-expected.txt: Added.
* compositing/shared-backing/overflow-scroll/shared-layer-composited-bounds.html: Added.
* compositing/shared-backing/overflow-scroll/shared-layer-nested-relative-stacking-expected.txt: Added.
* compositing/shared-backing/overflow-scroll/shared-layer-nested-relative-stacking.html: Added.
* compositing/shared-backing/overflow-scroll/shared-layer-repaint-expected.txt: Added.
* compositing/shared-backing/overflow-scroll/shared-layer-repaint.html: Added.
* compositing/shared-backing/partial-compositing-update-expected.txt: Added.
* compositing/shared-backing/partial-compositing-update.html: Added.
* compositing/shared-backing/partial-compositing-update2-expected.txt: Added.
* compositing/shared-backing/partial-compositing-update2.html: Added.
* compositing/shared-backing/remove-sharing-layer-expected.txt: Added.
* compositing/shared-backing/remove-sharing-layer.html: Added.
* compositing/shared-backing/sharing-cached-clip-rects-expected.txt: Added.
* compositing/shared-backing/sharing-cached-clip-rects.html: Added.
* platform/ios-wk2/TestExpectations:
* platform/ios-wk2/compositing/shared-backing/overflow-scroll/absolute-in-stacking-relative-in-scroller-expected.txt: Copied from LayoutTests/platform/ios/compositing/overflow/scrolling-content-clip-to-viewport-expected.txt.
* platform/ios-wk2/compositing/shared-backing/overflow-scroll/composited-absolute-in-absolute-in-relative-in-scroller-expected.txt: Added.
* platform/ios-wk2/compositing/shared-backing/overflow-scroll/nested-absolute-with-clipping-in-stacking-overflow-expected.txt: Added.
* platform/ios-wk2/compositing/shared-backing/overflow-scroll/previous-sibling-prevents-inclusiveness-expected.txt: Added.
* platform/ios-wk2/compositing/shared-backing/overflow-scroll/relative-in-clipping-in-scroller-in-clipping-expected.txt: Added.
* platform/ios-wk2/compositing/shared-backing/overflow-scroll/relative-in-clipping-in-scroller-in-relative-clipping-expected.txt: Added.
* platform/ios-wk2/compositing/shared-backing/overflow-scroll/relative-in-div-in-overflow-scroll-expected.txt: Copied from LayoutTests/platform/ios/compositing/overflow/scrolling-content-clip-to-viewport-expected.txt.
* platform/ios-wk2/compositing/shared-backing/overflow-scroll/scrolled-contents-has-painted-content-expected.txt: Copied from LayoutTests/platform/ios/compositing/overflow/scrolling-content-clip-to-viewport-expected.txt.
* platform/ios-wk2/scrollingcoordinator/scrolling-tree/nested-absolute-in-absolute-overflow-expected.txt:
* platform/ios-wk2/scrollingcoordinator/scrolling-tree/nested-absolute-in-relative-in-overflow-expected.txt:
* platform/ios-wk2/scrollingcoordinator/scrolling-tree/nested-absolute-in-sc-overflow-expected.txt:
* platform/ios-wk2/scrollingcoordinator/scrolling-tree/positioned-nodes-complex-expected.txt:
* platform/ios/compositing/geometry/limit-layer-bounds-clipping-ancestor-expected.txt:
* platform/ios/compositing/overflow/clipping-behaviour-change-is-not-propagated-to-descendants-expected.txt:
* platform/ios/compositing/overflow/clipping-behaviour-change-is-not-propagated-to-descendants2-expected.txt:
* platform/ios/compositing/overflow/scrolling-content-clip-to-viewport-expected.txt:
* platform/mac-wk1/compositing/overflow/scrolling-content-clip-to-viewport-expected.txt:
* platform/mac-wk2/TestExpectations:
* platform/mac/compositing/overflow/clipping-behaviour-change-is-not-propagated-to-descendants-expected.txt:
* platform/mac/compositing/overflow/clipping-behaviour-change-is-not-propagated-to-descendants2-expected.txt:
* scrollingcoordinator/scrolling-tree/nested-absolute-in-absolute-overflow-expected.txt:
* scrollingcoordinator/scrolling-tree/nested-absolute-in-absolute-overflow.html:
* scrollingcoordinator/scrolling-tree/nested-absolute-in-relative-in-overflow-expected.txt:
* scrollingcoordinator/scrolling-tree/nested-absolute-in-relative-in-overflow.html:
* scrollingcoordinator/scrolling-tree/nested-absolute-in-sc-overflow-expected.txt:
* scrollingcoordinator/scrolling-tree/nested-absolute-in-sc-overflow.html:
* scrollingcoordinator/scrolling-tree/positioned-nodes-complex-expected.txt:
* scrollingcoordinator/scrolling-tree/positioned-nodes-complex.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245170 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoWinCairo WebKitTestRunner should take platform/wk2/TestExpectations into account
Hironori.Fujii@sony.com [Fri, 10 May 2019 02:38:33 +0000 (02:38 +0000)]
WinCairo WebKitTestRunner should take platform/wk2/TestExpectations into account
https://bugs.webkit.org/show_bug.cgi?id=197734

Reviewed by Don Olmstead.

* Scripts/webkitpy/port/win.py:
(WinCairoPort._search_paths): Add 'wk2' and
additional_platform_directory to paths as well as other ports do.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245169 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoFix crashes related to pointer authentication for primitive gigacage
keith_miller@apple.com [Fri, 10 May 2019 02:02:31 +0000 (02:02 +0000)]
Fix crashes related to pointer authentication for primitive gigacage
https://bugs.webkit.org/show_bug.cgi?id=197763
<rdar://problem/50629257>

Reviewed by Saam Barati.

This fixes two bugs related to PAC for caging. The first is that
we didn't clear the high bits of the size register going into the
patchpoint to tag the new buffer for NewArrayBuffer. The second is
that the GC needs to strip all stack pointers when considering
them as a conservative root.

* ftl/FTLLowerDFGToB3.cpp:
(JSC::FTL::DFG::LowerDFGToB3::compileNewTypedArray):
* heap/ConservativeRoots.cpp:
(JSC::ConservativeRoots::genericAddPointer):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245168 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoWeb Inspector: Uncaught Exception: null is not an object (evaluating 'resource.parent...
drousso@apple.com [Fri, 10 May 2019 00:32:09 +0000 (00:32 +0000)]
Web Inspector: Uncaught Exception: null is not an object (evaluating 'resource.parentFrame.securityOrigin')
https://bugs.webkit.org/show_bug.cgi?id=196572
<rdar://problem/49578338>

Reviewed by Timothy Hatcher.

* UserInterface/Views/SourcesNavigationSidebarPanel.js:
(WI.SourcesNavigationSidebarPanel.prototype._addResource):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245167 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoWeb Inspector: REGRESSION: unable to expand/collapse non-selectable WI.TreeOutline
drousso@apple.com [Fri, 10 May 2019 00:15:13 +0000 (00:15 +0000)]
Web Inspector: REGRESSION: unable to expand/collapse non-selectable WI.TreeOutline
https://bugs.webkit.org/show_bug.cgi?id=197591

Reviewed by Timothy Hatcher.

* UserInterface/Views/TreeElement.js:
(WI.TreeElement.treeElementToggled):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245166 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoWeb Inspector: Uncaught Exception: null is not an object (evaluating 'url.startsWith')
drousso@apple.com [Fri, 10 May 2019 00:12:31 +0000 (00:12 +0000)]
Web Inspector: Uncaught Exception: null is not an object (evaluating 'url.startsWith')
https://bugs.webkit.org/show_bug.cgi?id=196662
<rdar://problem/49659633>

Reviewed by Timothy Hatcher.

* UserInterface/Views/ConsoleMessageView.js:
(WI.ConsoleMessageView.prototype._appendLocationLink):
(WI.ConsoleMessageView.prototype._linkifyLocation): Deleted.

* UserInterface/Base/Main.js:
(WI.linkifyLocation):
(WI.linkifySourceCode): Added.
Split the logic into two functions, as there's no need to re-fetch the `SourceCode` from a
given url if you already have the `SourceCode` in the caller.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245165 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoUnreviewed, rolling out most of r245151.
achristensen@apple.com [Thu, 9 May 2019 23:29:41 +0000 (23:29 +0000)]
Unreviewed, rolling out most of r245151.

Caused assertions in High Sierra

Reverted most of changeset:

"Remove now-unnecessary Connection::sendMessageWithReply"
https://bugs.webkit.org/show_bug.cgi?id=197747
https://trac.webkit.org/changeset/245151

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245164 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoUnreview, fix test to have a try-catch.
keith_miller@apple.com [Thu, 9 May 2019 23:05:44 +0000 (23:05 +0000)]
Unreview, fix test to have a try-catch.

* stress/many-nested-functions-parser-stack-overflow.js:
(catch):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245163 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[iOS] Right command key has wrong value for property code
dbates@webkit.org [Thu, 9 May 2019 23:04:40 +0000 (23:04 +0000)]
[iOS] Right command key has wrong value for property code
https://bugs.webkit.org/show_bug.cgi?id=193876
<rdar://problem/47577308>

Reviewed by Brent Fulgham.

Source/WebCore:

We're looking for the wrong Windows virtual key code for the right command key.
Substitute VK_APPS for VK_RWIN so we can identify the right command key and return
the correct value for the code property of the DOM key event.

* platform/ios/PlatformEventFactoryIOS.mm:
(WebCore::codeForKeyEvent):

LayoutTests:

Update test result.

* fast/events/ios/keydown-keyup-special-keys-in-non-editable-element-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245162 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[iOS] Numpad comma key has incorrect keyIdentifier property
dbates@webkit.org [Thu, 9 May 2019 22:59:36 +0000 (22:59 +0000)]
[iOS] Numpad comma key has incorrect keyIdentifier property
https://bugs.webkit.org/show_bug.cgi?id=197753
<rdar://problem/50636274>

Reviewed by Brent Fulgham.

Source/WebCore:

Map the Numpad Comma HID usage code to its Window virtual key code, VK_SEPARATOR,
so that can compute the keyIdentifier property for the Numpad Comma key. Also
consider this key as a keypad key just like we do on Mac. There is no discernable
difference for doing so on iOS because the non-keypad code path computes the same result
due to GraphicsServices having fixed up the input string for the Numpad Comma to be
",", which is want.

* platform/ios/KeyEventIOS.mm:
(WebCore::windowsKeyCodeForKeyCode): Map kHIDUsage_KeypadComma to VK_SEPARATOR.
* platform/ios/PlatformEventFactoryIOS.mm:
(WebCore::codeForKeyEvent): Add a comment to explain that this key is only on
JIS keyboards.
(WebCore::isKeypadEvent): Return true for the Numpad Comma key.

Source/WebCore/PAL:

Expose enumerator for Numpad Comma.

* pal/spi/cocoa/IOKitSPI.h:

Tools:

Recognize "numpadComma" and return the appropriate HID usage code.

* WebKitTestRunner/ios/HIDEventGenerator.mm:
(hidUsageCodeForCharacter):

LayoutTests:

Update test results.

* fast/events/ios/keydown-keyup-special-keys-in-non-editable-element-expected.txt:
* fast/events/ios/keydown-keyup-special-keys-in-non-editable-element.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245161 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoDo not mix inline and block level boxes.
zalan@apple.com [Thu, 9 May 2019 21:37:56 +0000 (21:37 +0000)]
Do not mix inline and block level boxes.
https://bugs.webkit.org/show_bug.cgi?id=197462
<rdar://problem/50369362>

Reviewed by Antti Koivisto.

Source/WebCore:

This patch tightens the remove-anonymous-wrappers logic by checking if the removal would
produce an inline-block sibling mix.
When a block level box is removed from the tree, we check if after the removal the anonymous sibling block
boxes are still needed or whether we can removed them as well (and have only inline level child boxes).
In addition to checking if the container is anonymous and is part of a continuation, we also need to check
if collapsing it (and by that moving its children one level up) would cause a inline-block box mix.

Test: fast/ruby/continuation-and-column-spanner-crash.html

* rendering/updating/RenderTreeBuilder.cpp:
(WebCore::RenderTreeBuilder::removeAnonymousWrappersForInlineChildrenIfNeeded):
* rendering/updating/RenderTreeBuilderContinuation.cpp:
(WebCore::RenderTreeBuilder::Continuation::cleanupOnDestroy):

LayoutTests:

* fast/ruby/continuation-and-column-spanner-crash-expected.txt: Added.
* fast/ruby/continuation-and-column-spanner-crash.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245158 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoUnreviewed, rebaseline test that is now passing after r245088.
ryanhaddad@apple.com [Thu, 9 May 2019 20:34:02 +0000 (20:34 +0000)]
Unreviewed, rebaseline test that is now passing after r245088.

* platform/ios-simulator/webgl/2.0.0/conformance/glsl/misc/shaders-with-invariance-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245157 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoRefine AudioSession route sharing policy
eric.carlson@apple.com [Thu, 9 May 2019 20:32:55 +0000 (20:32 +0000)]
Refine AudioSession route sharing policy
https://bugs.webkit.org/show_bug.cgi?id=197742
<rdar://problem/50590818>

Reviewed by Darin Adler.

Source/WebCore:

No new tests, updated AVAudioSessionRouteSharingPolicy API test.

* platform/audio/cocoa/MediaSessionManagerCocoa.mm:
(MediaSessionManagerCocoa::updateSessionState):

Tools:

* TestWebKitAPI/Tests/WebKitLegacy/ios/AudioSessionCategoryIOS.mm:
(TestWebKitAPI::TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245156 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[ANGLE] Update ANGLE
don.olmstead@sony.com [Thu, 9 May 2019 20:27:21 +0000 (20:27 +0000)]
[ANGLE] Update ANGLE
https://bugs.webkit.org/show_bug.cgi?id=197676
<rdar://problem/50609501>

Reviewed by Alex Christensen.

* changes.diff:
Updated with all the previous fixes.
* include/EGL/eglplatform.h:
Use the more generic name USE_SYSTEM_EGL not USE_WPE.
* src/common/debug.h:
Fix a stray COMPILER_GCC to be __GNUC__.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245155 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[iOS] Lazily request keyboard on first hardware keydown when a non-editable element...
dbates@webkit.org [Thu, 9 May 2019 20:18:44 +0000 (20:18 +0000)]
[iOS] Lazily request keyboard on first hardware keydown when a non-editable element is focused
https://bugs.webkit.org/show_bug.cgi?id=197746
<rdar://problem/50630406>

Reviewed by Brent Fulgham.

Be even lazier. We only need to create a keyboard if an editable element is focused or on
the first hardware keydown when a non-editable element is focused. We currently do the former.
Let's also do the latter.

* UIProcess/ios/WKContentViewInteraction.h:
* UIProcess/ios/WKContentViewInteraction.mm:
(-[WKContentView cleanupInteraction]): Reset bookkeeping.
(-[WKContentView _requiresKeyboardWhenFirstResponder]): Further condition the result on
whether we have seen a hardware keydown event in a non-editable element.
(-[WKContentView _handleKeyUIEvent:]): If this is the first hardware key event we've seen then
update state and call -reloadInputView. -reloadInputView will call us back in -_requiresKeyboardWhenFirstResponder
and we will now return YES. So, it will create the keyboard.
(-[WKContentView _hardwareKeyboardAvailabilityChanged]): Reset bookkeeping.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245154 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoGuard long press link preview with a USE macro rather than complicated includes
dino@apple.com [Thu, 9 May 2019 19:42:44 +0000 (19:42 +0000)]
Guard long press link preview with a USE macro rather than complicated includes
https://bugs.webkit.org/show_bug.cgi?id=197728
<rdar://problem/50604700>

Reviewed by Wenson Hsieh.

It was getting complicated to manage the coordination between WebKitAdditions
and WebKit without having to split up the includes into multiple files
or have lots of duplication.

Instead, WebKitAdditions will now define a USE macro which can be used
in various places through WebKit.

* UIProcess/ios/WKContentViewInteraction.h:
* UIProcess/ios/WKContentViewInteraction.mm:
(-[WKContentView dragInteraction:itemsForBeginningSession:]):
(-[WKContentView _contentsOfUserInterfaceItem:]):
(previewIdentifierForElementAction):
(-[WKContentView shouldUsePreviewForLongPress]):
(-[WKContentView _registerPreview]):
(-[WKContentView _unregisterPreview]):
(shouldUsePreviewForLongPress): Deleted.
(-[WKContentView _registerPreviewLongPress]): Deleted.
(-[WKContentView _unregisterPreviewLongPress]): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245153 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoparseStatementListItem needs a stack overflow check
keith_miller@apple.com [Thu, 9 May 2019 19:40:42 +0000 (19:40 +0000)]
parseStatementListItem needs a stack overflow check
https://bugs.webkit.org/show_bug.cgi?id=197749
JSTests:

Reviewed by Saam Barati.

* stress/many-nested-functions-parser-stack-overflow.js: Added.

Source/JavaScriptCore:

<rdar://problem/50302697>

Reviewed by Saam Barati.

There currently exists a path in the parser where you can loop
arbibrarily many times without a stack overflow check. This patch
adds a check to parseStatementListItem to break that cycle.

* parser/Parser.cpp:
(JSC::Parser<LexerType>::parseStatementListItem):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245152 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoRemove now-unnecessary Connection::sendMessageWithReply
achristensen@apple.com [Thu, 9 May 2019 19:40:21 +0000 (19:40 +0000)]
Remove now-unnecessary Connection::sendMessageWithReply
https://bugs.webkit.org/show_bug.cgi?id=197747

Reviewed by Tim Horton.

The WebProcess messages that use it already do the right thing with CompletionHandlers and can be replaced by sendWithAsyncReply.
The SecItemShim message that uses it can be replaced by sendSync.

This patch only increases responsiveness when clearing website data and removes the need for a WorkQueue dedicated entirely to the SecItemShim.

* Platform/IPC/Connection.cpp:
(IPC::Connection::invalidate):
(IPC::Connection::processIncomingSyncReply):
(IPC::Connection::connectionDidClose):
(IPC::Connection::sendMessageWithReply): Deleted.
* Platform/IPC/Connection.h:
(IPC::Connection::send):
(IPC::Connection::sendWithReply): Deleted.
* Shared/mac/SecItemResponseData.h:
* Shared/mac/SecItemShim.cpp:
(WebKit::sendSecItemRequest):
(WebKit::workQueue): Deleted.
* UIProcess/WebProcessProxy.cpp:
(WebKit::WebProcessProxy::fetchWebsiteData):
(WebKit::WebProcessProxy::deleteWebsiteData):
(WebKit::WebProcessProxy::deleteWebsiteDataForOrigins):
* WebProcess/WebProcess.messages.in:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245151 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoBuild fix for iOS.
dino@apple.com [Thu, 9 May 2019 19:38:58 +0000 (19:38 +0000)]
Build fix for iOS.

* UIProcess/WKImagePreviewViewController.mm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245150 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoRemove unnecessary test SPI after r244883
achristensen@apple.com [Thu, 9 May 2019 19:35:06 +0000 (19:35 +0000)]
Remove unnecessary test SPI after r244883
https://bugs.webkit.org/show_bug.cgi?id=197575

Reviewed by Tim Horton.

Source/WebKit:

We now have two ways of doing the same thing.  Let's remove the one that's only for testing,
and make the tests clean up like an application would.

* NetworkProcess/NetworkProcess.cpp:
(WebKit::NetworkProcess::removeCredential): Deleted.
* NetworkProcess/NetworkProcess.h:
* NetworkProcess/NetworkProcess.messages.in:
* NetworkProcess/cocoa/NetworkProcessCocoa.mm:
(WebKit::NetworkProcess::removeCredential): Deleted.
* UIProcess/API/Cocoa/WKProcessPool.mm:
(-[WKProcessPool _removeCredential:forProtectionSpace:completionHandler:]): Deleted.
* UIProcess/API/Cocoa/WKProcessPoolPrivate.h:
* UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::removeCredential): Deleted.
* UIProcess/WebProcessPool.h:

Tools:

* TestWebKitAPI/Tests/WebKitCocoa/Challenge.mm:
(TEST):
* TestWebKitAPI/Tests/WebKitCocoa/WKWebsiteDatastore.mm:
(TestWebKitAPI::TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245149 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoREGRESSION (r241734): 1% slower PLT on iPad
dbates@webkit.org [Thu, 9 May 2019 18:58:45 +0000 (18:58 +0000)]
REGRESSION (r241734): 1% slower PLT on iPad
https://bugs.webkit.org/show_bug.cgi?id=197745
<rdar://problem/50457731>

Reviewed by Per Arne Vollan.

For now, only create a keyboard when WKContentView becomes first responder if a hardware keyboard
is attached or an editable element is focused to recover the 1% loss when a keyboard is not attached.
We can do better and by that I mean be lazier. We'll do this <https://bugs.webkit.org/show_bug.cgi?id=197746>.

In r241734 we unified the key event handling code paths so we use exactly one for both software and
hardware key events. We took a simple approach of always requesting UIKit to create a keyboard when
the WKContentView becomes first responder. We did this so that we could continue listening for hardware
key events even when a non-editable element is focused and dispatch DOM events. As it turns out, always
creating a keyboard is expensive and caused a ~1% slowdown in page load time on iPad.

* UIProcess/ios/WKContentViewInteraction.mm:
(-[WKContentView _requiresKeyboardWhenFirstResponder]): Only create a keyboard if a hardware keyboard is
attached or an editable element is focused.
(-[WKContentView _hardwareKeyboardAvailabilityChanged]): Reload all input view (this will cause keyboard
creation) if -_requiresKeyboardWhenFirstResponder returns YES.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245148 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agofast/hidpi/video-controls-in-hidpi.html sometimes asserts in WK1
simon.fraser@apple.com [Thu, 9 May 2019 18:26:24 +0000 (18:26 +0000)]
fast/hidpi/video-controls-in-hidpi.html sometimes asserts in WK1
https://bugs.webkit.org/show_bug.cgi?id=197695

Reviewed by Zalan Bujtas.

With the backing-sharing changes that landed in r245058, some WebKit1 tests with media controls asserted in
RenderLayerBacking::computeParentGraphicsLayerRect() because a layer would have a m_ancestorClippingLayer,
but backgroundClipRect() would return an infinite rect.

This happened when a layer tree change caused the layer's compositing ancestor to no longer isolate
composited blending (which affects the behavior of RenderLayerCompositor:clippedByAncestor()), but we failed
to mark its composited children as needing the configuration update which would eliminate their m_ancestorClippingLayers.

The fix is to call setChildrenNeedCompositingGeometryUpdate() when isolatesCompositedBlending changes. We don't haev
setChildrenNeedCompositingConfigurationUpdate(), but setChildrenNeedCompositingGeometryUpdate() has the desired side-effect.

I was unable to make a standalone test case for this, but the code is exercised by media control tests.

* rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::computeCompositingRequirements):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245147 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoREGRESSION (r245064): ASSERTION FAILED: m_ptr seen with wasm.yaml/wasm/js-api/test_Da...
keith_miller@apple.com [Thu, 9 May 2019 17:56:47 +0000 (17:56 +0000)]
REGRESSION (r245064): ASSERTION FAILED: m_ptr seen with wasm.yaml/wasm/js-api/test_Data.js.wasm-slow-memory
https://bugs.webkit.org/show_bug.cgi?id=197740

Reviewed by Saam Barati.

If a TypedArray constructor is called with just 0 as the first argument, we don't allocate a backing vector.
This means we need to handle null when calling vector() in ConstructionContext.

* runtime/JSArrayBufferView.h:
(JSC::JSArrayBufferView::ConstructionContext::vector const):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245145 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[iOS] Unable to commit search on MSN.com, qq.com, or sina.com.cn using enter key...
dbates@webkit.org [Thu, 9 May 2019 17:41:20 +0000 (17:41 +0000)]
[iOS] Unable to commit search on MSN.com, qq.com, or sina.com.cn using enter key (hardware or software keyboard)
https://bugs.webkit.org/show_bug.cgi?id=197632
<rdar://problem/47902054>

Reviewed by Brent Fulgham.

Source/WebKit:

Fixes an issue where it is not possible to submit a <form> with target = "_blank": a form that
opens a new window.

By default we only allow popups to open if they were user initiated (like when a person clicks
on a link). We achieve this by putting a token on the stack, called the UserGestureToken when
WebCore processes an event from WebKit. So long as this token is on the stack we consider
all requests to open a popup to be user initiated. And we implicitly submit a form when pressing
the Return key in an HTML input element during the processing of a TextInputEvent dispatched as
part of inserting a '\n' into the field. On Mac, the keydown dispatches a TextInputEvent synchronously.
However on iOS text insertion, and hence a dispatch of a TextInputEvent event, occurs asynchronously
with respect to the keydown event. So, by the time the UI process calls back to the WebProcess
to perform the text insertion of '\n' we have long since popped the UserGestureToken off the stack
and hence we disallow opening a popup. To fix this, when -insertText is called we query the keyboard
to determine if it's being called by the keyboard. If it is then we can assume that this is
part of key event handling and hence was initiated by the user. We can pass along this detail
to the WebProcess for it to push a new UserGestureToken onto the stack.

For now we only track whether text inserted by the keyboard was user initiated or not. In
<https://bugs.webkit.org/show_bug.cgi?id=197721> we will fix this up for all editing commands.

* Platform/spi/ios/UIKitSPI.h: Expose SPI.
* Shared/Cocoa/InsertTextOptions.cpp:
(IPC::ArgumentCoder<WebKit::InsertTextOptions>::encode):
(IPC::ArgumentCoder<WebKit::InsertTextOptions>::decode):
Encode and decode whether we are processing a user gesture.

* Shared/Cocoa/InsertTextOptions.h:
* UIProcess/ios/WKContentViewInteraction.mm:
(-[WKContentView insertText:]): Query the keyboard to determine whether it called us or
the embedding client did. We only want to privilege user initiated actions (the keyboard).
* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::insertTextAsync): Push a UserGestureToken onto the stack that is initialized
depending on whether we are or are not processing a user gesture.

LayoutTests:

Add tests to ensure we fire input and keypress events in the correct order and that we can
submit a <form> with target = "_blank" using the Return key.

* fast/events/ios/fire-input-and-keypress-on-return-key-expected.txt: Added.
* fast/events/ios/fire-input-and-keypress-on-return-key.html: Added.
* fast/events/ios/submit-form-target-blank-using-return-key-expected.txt: Added.
* fast/events/ios/submit-form-target-blank-using-return-key.html: Added.
* platform/ios/TestExpectations: Skip the test until we have the UIKit SPI added
in <rdar://problem/50596032>.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245144 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[ews-app] Add migrations file to repository
aakash_jain@apple.com [Thu, 9 May 2019 17:26:11 +0000 (17:26 +0000)]
[ews-app] Add migrations file to repository
https://bugs.webkit.org/show_bug.cgi?id=197729

Reviewed by Jonathan Bedard.

* BuildSlaveSupport/ews-app/ews/migrations/0001_initial.py: Added. Auto-generated by Django based on models' information.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245143 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoCorrect delayed load event handling
bfulgham@apple.com [Thu, 9 May 2019 17:14:06 +0000 (17:14 +0000)]
Correct delayed load event handling
https://bugs.webkit.org/show_bug.cgi?id=197679
<rdar://problem/50423334>

Reviewed by Alex Christensen.

We need to properly account for the fact that JavaScript might run
while performing loads.

* dom/Document.cpp:
(WebCore::Document::loadEventDelayTimerFired):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245142 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[ews-build] Fix formatting issues and typos
aakash_jain@apple.com [Thu, 9 May 2019 17:13:47 +0000 (17:13 +0000)]
[ews-build] Fix formatting issues and typos
https://bugs.webkit.org/show_bug.cgi?id=197737

Reviewed by Jonathan Bedard.

* BuildSlaveSupport/ews-build/steps.py:
(CompileWebKit.evaluateCommand): Removed extra empty line.
* BuildSlaveSupport/ews-build/steps_unittest.py:
(TestRunWebKitPerlTests): Fixed typo.
* BuildSlaveSupport/ews-app/ews/models/step.py:
(Step.save_step): Changed log level to info so that it is logged in production.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245141 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoFix 32-bit watchOS engineering builds after r244726.
aestes@apple.com [Thu, 9 May 2019 16:57:25 +0000 (16:57 +0000)]
Fix 32-bit watchOS engineering builds after r244726.

Unreviewed.

* Configurations/DebugRelease.xcconfig:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245140 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoVersioning.
kocsen_chung@apple.com [Thu, 9 May 2019 16:45:46 +0000 (16:45 +0000)]
Versioning.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245139 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[ews-app] Production and Development env should configure DEBUG appropriately
aakash_jain@apple.com [Thu, 9 May 2019 16:09:35 +0000 (16:09 +0000)]
[ews-app] Production and Development env should configure DEBUG appropriately
https://bugs.webkit.org/show_bug.cgi?id=197700

Reviewed by Jonathan Bedard.

* BuildSlaveSupport/ews-app/ews-app/settings.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245136 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[Win10] Some tests are failing only on specific machines
pvollan@apple.com [Thu, 9 May 2019 14:42:35 +0000 (14:42 +0000)]
[Win10] Some tests are failing only on specific machines
https://bugs.webkit.org/show_bug.cgi?id=197509

Unreviewed test gardening.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245135 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agopointerevents/ios/touch-action-none-in-overflow-scrolling-touch.html is a timeout
graouts@webkit.org [Thu, 9 May 2019 14:16:44 +0000 (14:16 +0000)]
pointerevents/ios/touch-action-none-in-overflow-scrolling-touch.html is a timeout
https://bugs.webkit.org/show_bug.cgi?id=197738
<rdar://problem/50588613>

Reviewed by Antti Koivisto.

Source/WebKit:

We clear the touch actions for a given touch identifier when the matching touch is released in -[WKContentViewInteraction _handleTouchActionsForTouchEvent:].
This happens before -[WKScrollingNodeScrollViewDelegate scrollViewWillEndDragging:withVelocity:targetContentOffset:] is called and thus, which we correctly
respected the touch actions during the panning gesture, we would not be able to respect them during the deceleration animation. This caused the test to fail
because it would check that no scrolling happened after the panning gesture completed, ie. during the deceleration animation.

To work around this, we now store the touch actions for a given UIScrollView interaction in the ScrollingTreeScrollingNodeDelegateIOS object when the interaction
starts.

* UIProcess/RemoteLayerTree/ios/ScrollingTreeScrollingNodeDelegateIOS.h:
(WebKit::ScrollingTreeScrollingNodeDelegateIOS::activeTouchActions const):
(WebKit::ScrollingTreeScrollingNodeDelegateIOS::clearActiveTouchActions):
* UIProcess/RemoteLayerTree/ios/ScrollingTreeScrollingNodeDelegateIOS.mm:
(-[WKScrollingNodeScrollViewDelegate scrollViewWillEndDragging:withVelocity:targetContentOffset:]):
(-[WKScrollingNodeScrollViewDelegate _scrollView:adjustedOffsetForOffset:translation:startPoint:locationInView:horizontalVelocity:verticalVelocity:]):
(WebKit::ScrollingTreeScrollingNodeDelegateIOS::computeActiveTouchActionsForGestureRecognizer):
(WebKit::ScrollingTreeScrollingNodeDelegateIOS::activeTouchActionsForGestureRecognizer const): Deleted.

LayoutTests:

To pass, this test must also disable the legacy "-webkit-overflow-scrolling: touch" behavior.

* pointerevents/ios/touch-action-none-in-overflow-scrolling-touch.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245134 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[CMake] Detect SSE2 at compile time
commit-queue@webkit.org [Thu, 9 May 2019 10:43:43 +0000 (10:43 +0000)]
[CMake] Detect SSE2 at compile time
https://bugs.webkit.org/show_bug.cgi?id=196488

Patch by Xan López <xan@igalia.com> on 2019-05-09
Reviewed by Carlos Garcia Campos.

.:

* Source/cmake/DetectSSE2.cmake: Added.
* Source/cmake/WebKitCompilerFlags.cmake: Detect SSE2 support and
add SSE2 to the global compiler flags.

Source/JavaScriptCore:

* assembler/MacroAssemblerX86Common.cpp: Remove unnecessary (and
incorrect) static_assert.
(JSC::MacroAssemblerX86Common::collectCPUFeatures):
* assembler/MacroAssemblerX86Common.h: Remove SSE2 flags.

Tools:

* Scripts/webkitdirs.pm:
(generateBuildSystemFromCMakeProject): Do not add SSE2 flags here
for x86 builds. This is now handled in WebKitCompilerFlags.cmake.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245127 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoElements with "display: inline-block" don't have a touch-action region
antti@apple.com [Thu, 9 May 2019 09:41:37 +0000 (09:41 +0000)]
Elements with "display: inline-block" don't have a touch-action region
https://bugs.webkit.org/show_bug.cgi?id=197281
<rdar://problem/50535081>

Reviewed by Antoine Quint.

Source/WebCore:

Test: pointerevents/ios/touch-action-region-inline-block.html

PaintPhase::EventRegion failed to traverse into inline boxes.

* rendering/InlineElementBox.cpp:
(WebCore::InlineElementBox::paint):
* rendering/RenderElement.cpp:
(WebCore::RenderElement::paintAsInlineBlock):

LayoutTests:

* pointerevents/ios/touch-action-region-inline-block-expected.txt: Added.
* pointerevents/ios/touch-action-region-inline-block.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245126 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[iOS] Correctly handle overlapping regions for elements with a touch-action property
graouts@webkit.org [Thu, 9 May 2019 08:34:23 +0000 (08:34 +0000)]
[iOS] Correctly handle overlapping regions for elements with a touch-action property
https://bugs.webkit.org/show_bug.cgi?id=194813
<rdar://problem/48194708>

Reviewed by Antti Koivisto.

Source/WebKit:

We now use WebKit::touchActionsForPoint() to determine the touch actions for a given touch using its location in -[WKContentViewInteraction
_handleTouchActionsForTouchEvent:]. We then record these touch actions for the touch's identifier on the RemoteScrollingCoordinatorProxy.

Then, as we interact with a UIScrollView, we get its gesture recognizer and get its active touch identifier through the new
-[WKContentViewInteraction activeTouchIdentifierForGestureRecognizer:] method, and query the RemoteScrollingCoordinatorProxy for the touch
actions matching that touch identifier.

Tests: pointerevents/ios/touch-action-none-overlap.html
       pointerevents/ios/touch-action-pan-x-overlap.html
       pointerevents/ios/touch-action-pan-y-overlap.html
       pointerevents/ios/touch-action-pinch-zoom-overlap.html

* UIProcess/PageClient.h:
(WebKit::PageClient::activeTouchIdentifierForGestureRecognizer):
* UIProcess/RemoteLayerTree/RemoteScrollingCoordinatorProxy.cpp: Maintain a touch identifier to touch actions map.
(WebKit::RemoteScrollingCoordinatorProxy::activeTouchActionsForTouchIdentifier const):
(WebKit::RemoteScrollingCoordinatorProxy::setTouchActionsForTouchIdentifier):
(WebKit::RemoteScrollingCoordinatorProxy::clearTouchActionsForTouchIdentifier):
(WebKit::RemoteScrollingCoordinatorProxy::touchActionDataAtPoint const): Deleted.
(WebKit::RemoteScrollingCoordinatorProxy::touchActionDataForScrollNodeID const): Deleted.
(WebKit::RemoteScrollingCoordinatorProxy::setTouchDataForTouchIdentifier): Deleted.
(WebKit::RemoteScrollingCoordinatorProxy::clearTouchDataForTouchIdentifier): Deleted.
* UIProcess/RemoteLayerTree/RemoteScrollingCoordinatorProxy.h:
* UIProcess/RemoteLayerTree/ios/ScrollingTreeScrollingNodeDelegateIOS.h:
* UIProcess/RemoteLayerTree/ios/ScrollingTreeScrollingNodeDelegateIOS.mm:
(-[WKScrollingNodeScrollViewDelegate scrollViewWillEndDragging:withVelocity:targetContentOffset:]):
(-[WKScrollingNodeScrollViewDelegate _scrollView:adjustedOffsetForOffset:translation:startPoint:locationInView:horizontalVelocity:verticalVelocity:]):
(WebKit::ScrollingTreeScrollingNodeDelegateIOS::activeTouchActionsForGestureRecognizer const):
(WebKit::ScrollingTreeScrollingNodeDelegateIOS::touchActionData const): Deleted.
* UIProcess/ios/PageClientImplIOS.h:
* UIProcess/ios/PageClientImplIOS.mm:
(WebKit::PageClientImpl::activeTouchIdentifierForGestureRecognizer):
* UIProcess/ios/WKContentViewInteraction.h:
* UIProcess/ios/WKContentViewInteraction.mm:
(-[WKContentView activeTouchIdentifierForGestureRecognizer:]):
(-[WKContentView _handleTouchActionsForTouchEvent:]):

LayoutTests:

Add new tests that check that an element overlapping another element with a "touch-action" property is not affected
by the property set on that underlying element.

* pointerevents/ios/touch-action-none-overlap-expected.txt: Added.
* pointerevents/ios/touch-action-none-overlap.html: Added.
* pointerevents/ios/touch-action-pan-x-overlap-expected.txt: Added.
* pointerevents/ios/touch-action-pan-x-overlap.html: Added.
* pointerevents/ios/touch-action-pan-y-overlap-expected.txt: Added.
* pointerevents/ios/touch-action-pan-y-overlap.html: Added.
* pointerevents/ios/touch-action-pinch-zoom-overlap-expected.txt: Added.
* pointerevents/ios/touch-action-pinch-zoom-overlap.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245112 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[GLib] Rework WPE RunLoopSourcePriority values
zandobersek@gmail.com [Thu, 9 May 2019 06:37:11 +0000 (06:37 +0000)]
[GLib] Rework WPE RunLoopSourcePriority values
https://bugs.webkit.org/show_bug.cgi?id=197167

Reviewed by Carlos Garcia Campos.

The GLib priorities for the WPE port were initially aligned on the -70
value, theory being that this would help avoid any default-priority
GSources spawned in different dependency libraries affecting our
scheduling. Today it seems that extra caution might not be really
necessary.

This change aligns the base priority value with GLib's default priority
value of 0. We maintain the relativity of priority values by effectively
increasing each priority by 70.

* wtf/glib/RunLoopSourcePriority.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245096 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[WPE] Update dependencies to use libwpe + WPEBackend-fdo 1.3.0
aperez@igalia.com [Thu, 9 May 2019 06:30:39 +0000 (06:30 +0000)]
[WPE] Update dependencies to use libwpe + WPEBackend-fdo 1.3.0
https://bugs.webkit.org/show_bug.cgi?id=197697

Reviewed by Žan Doberšek.

* flatpak/org.webkit.WPEModules.yaml: Bump version and update checksums.
* wpe/jhbuild.modules: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245095 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoREGRESSION(r239915): [FreeType] White space skipped when rendering plain text with...
carlosgc@webkit.org [Thu, 9 May 2019 06:06:27 +0000 (06:06 +0000)]
REGRESSION(r239915): [FreeType] White space skipped when rendering plain text with noto CJK font
https://bugs.webkit.org/show_bug.cgi?id=197658

Reviewed by Michael Catanzaro.

Since r239915 we no longer overwrite control characters with zero width space, they are handled later when
filling the glyph pages. In Font::platformGlyphInit() there's an optimization to get the glyph of zero with
space character that assumes that control characters are always overwritten. Since the glyph for character at 0
index is always overwritten with zero width space, we can avoid loading the page for the actual zero width space
character and use the first page instead. In the particular case of noto CJK font, character at 0 is mapped to
the same glyph as space character, so space and zero width space end up being the same glyph. That breaks the
space width calculation, that returns 0 when isZeroWidthSpaceGlyph() is true. That's why spaces are no
longer rendered, ComplexTextController::adjustGlyphsAndAdvances() is setting the x advance for the space glyphs
to 0.

* platform/graphics/Font.cpp:
(WebCore::Font::platformGlyphInit): Use the actual zero width space page to get the glyph instead of 0 when
using FreeType.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245094 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoUnreviewed, build fix after r245064
ysuzuki@apple.com [Thu, 9 May 2019 05:41:03 +0000 (05:41 +0000)]
Unreviewed, build fix after r245064
https://bugs.webkit.org/show_bug.cgi?id=197110

* runtime/GenericTypedArrayView.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245093 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoFix internal and Windows builds.
achristensen@apple.com [Thu, 9 May 2019 05:06:52 +0000 (05:06 +0000)]
Fix internal and Windows builds.

* ANGLE.txt: Added.
Restore this file.  It was deleted, but it's needed.
* GLESv2.cmake:
It turns out CompositorNativeWindow11.cpp and Renderer11.cpp are needed, but the bot doesn't seem to have the right SDK.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245092 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoFix High Sierra and Windows builds.
achristensen@apple.com [Thu, 9 May 2019 04:55:22 +0000 (04:55 +0000)]
Fix High Sierra and Windows builds.

* GLESv2.cmake:
CompositorNativeWindow11.cpp was also having a hard time compiling on the bots.
* src/libANGLE/validationES3.cpp:
(gl::ValidateBindFragDataLocationIndexedEXT):
Remove a check for if an unsigned integer is less than 0.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245091 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoSource/ThirdParty/ANGLE:
achristensen@apple.com [Thu, 9 May 2019 04:49:11 +0000 (04:49 +0000)]
Source/ThirdParty/ANGLE:
Fix High Sierra and Windows builds.

* src/common/utilities.cpp:
(gl::priv::gLineModes):
* src/common/utilities.h:
(gl::IsLineMode):
Don't have a global constructor.  Use a function scoped static variable instead.
* GLESv2.cmake:
Renderer11.cpp was not building successfully on the bots.
I don't think it's needed, so just don't build it.

Source/WebCore:
Fix WPE build.

* CMakeLists.txt:
Bots wanted a "PUBLIC" or "PRIVATE" keyword here.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245090 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoSource/ThirdParty/ANGLE:
achristensen@apple.com [Thu, 9 May 2019 04:30:32 +0000 (04:30 +0000)]
Source/ThirdParty/ANGLE:
Fix High Sierra build.

* src/libANGLE/State.cpp:
(gl::State::kBufferSetters):
* src/libANGLE/State.h:
(gl::State::setBufferBinding):
Remove the need for a global constructor.

Source/WebCore:
Try to fix Linux build

* platform/graphics/ANGLEWebKitBridge.h:
Include headers consistently on all platforms.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245089 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoUpdate ANGLE
achristensen@apple.com [Thu, 9 May 2019 04:15:46 +0000 (04:15 +0000)]
Update ANGLE
https://bugs.webkit.org/show_bug.cgi?id=197676

Patch by Don Olmstead <don.olmstead@sony.com> on 2019-05-08
Reviewed by Alex Christensen.

Source/ThirdParty/ANGLE:

Many changed files.  Updated to trunk ANGLE 565441b1078b.

Source/WebCore:

* CMakeLists.txt:
* PlatformGTK.cmake:
* PlatformMac.cmake:
* PlatformPlayStation.cmake:
* PlatformWPE.cmake:
* PlatformWin.cmake:
* platform/graphics/GLContext.cpp:
* platform/graphics/egl/GLContextEGL.cpp:
* platform/graphics/opengl/Extensions3DOpenGLCommon.cpp:
* platform/graphics/opengl/Extensions3DOpenGLES.h:
* platform/graphics/opengl/TemporaryOpenGLSetting.cpp:
* platform/graphics/texmap/TextureMapperGLHeaders.h:

Tools:

* TestWebKitAPI/PlatformWin.cmake:
Remove unnecessary library linkage.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245088 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[Win] build-webkit should be ready for Visual Studio 2019
Hironori.Fujii@sony.com [Thu, 9 May 2019 02:06:39 +0000 (02:06 +0000)]
[Win] build-webkit should be ready for Visual Studio 2019
https://bugs.webkit.org/show_bug.cgi?id=196622

Reviewed by Ross Kirsling.

1. Accommodated to the new msbuild.exe location of VS 2019.
2. Removed the hard coded "Visual Studio 15" generator.

build-webkit is going to choose a Visual Studio generator version
matching with the latest installed VS. If you invoke vcvars.bat
before build-webkit, it chooses the corresponding Visual Studio
generator version by cheking 'VisualStudioVersion' env var which
is set by vcvars.bat.

* Scripts/webkitdirs.pm: Renamed $vcBuildPath to $msBuildPath. Removed $msBuildInstallDir.
(msBuildPath): When looking for MSBuild, look in
"MSBuild\Current\bin" (the new expected location) and fallback
to "MSBuild\15.0\bin" if we can't find an MSBuild.exe in the
first location.
(msBuildInstallDir): Renamed to msBuildPath.
(generateBuildSystemFromCMakeProject): Use -A switch to specify x64.
Add -G "Visual Studio " switch if a env var 'VisualStudioVersion'

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245087 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[iOS Debug] ASSERTION FAILED: !m_originalNode in WebCore::JSLazyEventListener::checkV...
cdumez@apple.com [Thu, 9 May 2019 01:34:05 +0000 (01:34 +0000)]
[iOS Debug] ASSERTION FAILED: !m_originalNode in WebCore::JSLazyEventListener::checkValidityForEventTarget(WebCore::EventTarget &)
https://bugs.webkit.org/show_bug.cgi?id=197696
<rdar://problem/50586956>

Reviewed by Simon Fraser.

Source/WebCore:

Setting the onorientationchange / onresize event handler on the body should set the event handler on the
window object, as per the HTML specification. However, calling body.addEventListener() with 'orientationchange'
or 'resize' should not set the event listener on the window object, only the body. Blink and Gecko seem to
behave as per specification but WebKit had a quirk for the addEventListener case. The quirk's implementation
is slightly wrong (because it is unsafe to take a JSLazyEventListener from a body element and add it to the
window, given that the JSLazyEventListener keeps a raw pointer to its element) and was causing crashes such
as <rdar://problem/24314027>. As a result, this patch simply drops the WebKit quirk, which will align our
behavior with other browsers and fix the crashes altogether.

Test: fast/events/ios/rotation/orientationchange-event-listener-on.body.html

* dom/Node.cpp:
(WebCore::tryAddEventListener):
(WebCore::tryRemoveEventListener):

LayoutTests:

Add layout test coverage.

* fast/events/ios/rotation/orientationchange-event-listener-on.body-expected.txt: Added.
* fast/events/ios/rotation/orientationchange-event-listener-on.body.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245086 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoUnreviewed, rolling out r245058.
ryanhaddad@apple.com [Thu, 9 May 2019 01:18:19 +0000 (01:18 +0000)]
Unreviewed, rolling out r245058.

Causes crashes under ASan / GuardMalloc

Reverted changeset:

"Implement backing-sharing in compositing layers, allowing
overlap layers to paint into the backing store of another
layer"
https://bugs.webkit.org/show_bug.cgi?id=197561
https://trac.webkit.org/changeset/245058

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245085 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months agoAccessGenerationState::emitExplicitExceptionHandler can clobber an in use register
sbarati@apple.com [Thu, 9 May 2019 00:49:35 +0000 (00:49 +0000)]
AccessGenerationState::emitExplicitExceptionHandler can clobber an in use register
https://bugs.webkit.org/show_bug.cgi?id=197715
<rdar://problem/50399252>

Reviewed by Filip Pizlo.

JSTests:

* stress/polymorphic-access-exception-handler-should-not-clobber-used-register.js: Added.
(foo):
(bar):

Source/JavaScriptCore:

AccessGenerationState::emitExplicitExceptionHandler was always clobbering
x86's r9 without considering if that register was needed to be preserved
by the IC. This leads to bad things when the DFG/FTL need that register when
OSR exitting after an exception from a GetById call.

* b3/air/AirCode.cpp:
(JSC::B3::Air::Code::Code):
* bytecode/PolymorphicAccess.cpp:
(JSC::AccessGenerationState::emitExplicitExceptionHandler):
* runtime/Options.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245084 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 months ago[PlayStation] Update port options
don.olmstead@sony.com [Thu, 9 May 2019 00:32:53 +0000 (00:32 +0000)]
[PlayStation] Update port options
https://bugs.webkit.org/show_bug.cgi?id=197723

Reviewed by Ross Kirsling.

.:

Update the options used.

* Source/cmake/OptionsPlayStation.cmake:

Tools:

ICU headers were missing from TestWTF.

* TestWebKitAPI/PlatformPlayStation.cmake:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@245083 268f45cc-cd09-0410-ab3c-d52691b4dbfc