Timeline
Mar 21, 2015:
- 9:55 PM Changeset in webkit [181832] by
-
- 37 edits1 copy4 adds in trunk
Remove the prefix for CSS Transforms
https://bugs.webkit.org/show_bug.cgi?id=142927
<rdar://problem/11568151>
Reviewed by Simon Fraser.
Source/WebCore:
Support the bare form of:
- transform
- transform-style
- transform-origin
- perspective
- perspective-origin
For transform-style, we also still differentiate the prefixed property
because we may want to use that as an indicator of legacy content.
This is mostly a find and replace, but with some related naming
changes. The DOM interface for WebKitCSSTransformValue remains
unchanged.
Tests: transforms/2d/transform-2d-prefixed.html
transforms/3d/general/prefixed-3dtransform-values.html
- bindings/js/JSCSSStyleDeclarationCustom.cpp:
(WebCore::propertyChangeMayRepaintNonDescendants):
- css/CSSComputedStyleDeclaration.cpp:
(WebCore::isLayoutDependent):
(WebCore::ComputedStyleExtractor::propertyValue):
- css/CSSParser.cpp:
(WebCore::isValidKeywordPropertyAndValue):
(WebCore::isKeywordPropertyID):
(WebCore::parseTranslateTransformValue):
(WebCore::CSSParser::parseValue):
(WebCore::CSSParser::parseTransformOrigin):
(WebCore::CSSParser::parsePerspectiveOrigin):
- css/CSSPropertyNames.in:
- css/StyleProperties.cpp:
(WebCore::StyleProperties::getPropertyValue):
(WebCore::StyleProperties::asText):
- css/WebKitCSSMatrix.cpp:
(WebCore::WebKitCSSMatrix::setMatrixValue):
- html/track/VTTCue.cpp:
(WebCore::VTTCueBox::applyCSSProperties):
- page/animation/AnimationController.cpp:
(WebCore::AnimationControllerPrivate::computeExtentOfAnimation):
- page/animation/CSSPropertyAnimation.cpp:
(WebCore::PropertyWrapperAcceleratedTransform::PropertyWrapperAcceleratedTransform):
(WebCore::CSSPropertyAnimationWrapperMap::CSSPropertyAnimationWrapperMap):
- page/animation/CompositeAnimation.cpp:
(WebCore::CompositeAnimation::computeExtentOfTransformAnimation):
- page/animation/ImplicitAnimation.cpp:
(WebCore::ImplicitAnimation::computeExtentOfTransformAnimation):
- page/animation/KeyframeAnimation.cpp:
(WebCore::KeyframeAnimation::computeExtentOfTransformAnimation):
(WebCore::KeyframeAnimation::validateTransformFunctionList):
- platform/graphics/GraphicsLayer.cpp:
(WebCore::GraphicsLayer::validateTransformOperations):
- platform/graphics/GraphicsLayerClient.h:
- platform/graphics/ca/GraphicsLayerCA.cpp:
(WebCore::propertyIdToString):
(WebCore::GraphicsLayerCA::moveOrCopyAnimations):
(WebCore::GraphicsLayerCA::addAnimation):
(WebCore::GraphicsLayerCA::isRunningTransformAnimation):
(WebCore::GraphicsLayerCA::createAnimationFromKeyframes):
(WebCore::GraphicsLayerCA::createTransformAnimationsFromKeyframes):
- platform/graphics/texmap/GraphicsLayerTextureMapper.cpp:
(WebCore::GraphicsLayerTextureMapper::addAnimation):
- platform/graphics/texmap/TextureMapperAnimation.cpp:
(WebCore::TextureMapperAnimation::applyInternal):
- platform/graphics/texmap/TextureMapperLayer.cpp:
(WebCore::TextureMapperLayer::syncAnimations):
- platform/graphics/texmap/coordinated/CoordinatedGraphicsLayer.cpp:
(WebCore::CoordinatedGraphicsLayer::selfOrAncestorHasActiveTransformAnimation):
(WebCore::CoordinatedGraphicsLayer::selfOrAncestorHaveNonAffineTransforms):
(WebCore::CoordinatedGraphicsLayer::addAnimation):
- rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::updateGeometry):
(WebCore::RenderLayerBacking::startAnimation):
(WebCore::RenderLayerBacking::startTransition):
(WebCore::RenderLayerBacking::graphicsLayerToCSSProperty):
(WebCore::RenderLayerBacking::cssToGraphicsLayerProperty):
- rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::requiresCompositingForAnimation):
(WebCore::RenderLayerCompositor::isRunningTransformAnimation):
- svg/SVGElement.cpp:
(WebCore::populateAttributeNameToCSSPropertyIDMap):
Source/WebKit2:
Rename AnimatedPropertyWebkitTransform to AnimatedPropertyTransform.
- Shared/CoordinatedGraphics/CoordinatedGraphicsArgumentCoders.cpp:
(IPC::ArgumentCoder<TextureMapperAnimation>::encode):
LayoutTests:
Convert the helper functions for transforms to use the unprefixed
form, and copy some tests to make sure the prefixed form still works.
- fast/css/getComputedStyle/computed-style-expected.txt:
- fast/css/getComputedStyle/computed-style-without-renderer-expected.txt:
- fast/css/getComputedStyle/resources/property-names.js:
- svg/css/getComputedStyle-basic-expected.txt:
- transforms/2d/computed-style-origin-expected.txt:
- transforms/2d/computed-style-origin.html:
- transforms/2d/resources/prefixed-transform-test-utils.js: Copied from LayoutTests/transforms/2d/resources/transform-test-utils.js.
(floatingPointEqual):
(compareMatrices):
(testTransforms):
- transforms/2d/resources/transform-test-utils.js:
(testTransforms):
- transforms/2d/transform-2d-prefixed-expected.txt: Added.
- transforms/2d/transform-2d-prefixed.html: Added.
- transforms/3d/general/prefixed-3dtransform-values-expected.txt: Added.
- transforms/3d/general/prefixed-3dtransform-values.html: Added.
- transitions/transition-end-event-multiple-04-expected.txt:
- transitions/transition-end-event-multiple-04.html:
- transitions/transition-end-event-transform-expected.txt:
- transitions/transition-end-event-transform.html:
- 9:08 PM Changeset in webkit [181831] by
-
- 11 edits in trunk
Fix a typo in Parser error message
https://bugs.webkit.org/show_bug.cgi?id=142942
Patch by Joseph Pecoraro <Joseph Pecoraro> on 2015-03-21
Reviewed by Alexey Proskuryakov.
Source/JavaScriptCore:
- jit/JITPropertyAccess.cpp:
(JSC::JIT::emitSlow_op_resolve_scope):
- jit/JITPropertyAccess32_64.cpp:
(JSC::JIT::emitSlow_op_resolve_scope):
- parser/Parser.cpp:
(JSC::Parser<LexerType>::parseClass):
Fix a common identifier typo.
LayoutTests:
- js/class-syntax-declaration-expected.txt:
- js/class-syntax-expression-expected.txt:
- js/class-syntax-super-expected.txt:
- js/script-tests/class-syntax-declaration.js:
- js/script-tests/class-syntax-expression.js:
- js/script-tests/class-syntax-super.js:
- 8:47 PM Changeset in webkit [181830] by
-
- 2 edits in trunk/LayoutTests
[Win] Try to get debug bots green.
- platform/win/TestExpectations:
- 8:14 PM Changeset in webkit [181829] by
-
- 7 edits in trunk
Computed Property names should allow only AssignmentExpressions not any Expression
https://bugs.webkit.org/show_bug.cgi?id=142902
Reviewed by Ryosuke Niwa.
Source/JavaScriptCore:
- parser/Parser.cpp:
(JSC::Parser<LexerType>::parseProperty):
Limit computed expressions to just assignment expressions instead of
any expression (which allowed comma expressions).
LayoutTests:
- js/basic-computed-property-name-expected.txt:
- js/object-literal-computed-methods-expected.txt:
- js/script-tests/basic-computed-property-name.js:
(runTest.runTest.runTest.runTest.runTest.runTest.runTest.runTestThrow):
- js/script-tests/object-literal-computed-methods.js:
Test that comma expressions are not allowed in computed property/method names.
- 5:07 PM Changeset in webkit [181828] by
-
- 3 edits in trunk/Source/JavaScriptCore
Make UnlinkedFunctionExecutable fit in a 128-byte cell.
<https://webkit.org/b/142939>
Reviewed by Mark Hahnenberg.
Re-arrange the members of UnlinkedFunctionExecutable so it can fit inside
a 128-byte heap cell instead of requiring a 256-byte one.
Threw in a static_assert to catch anyone pushing it over the limit again.
- bytecode/UnlinkedCodeBlock.cpp:
(JSC::UnlinkedFunctionExecutable::UnlinkedFunctionExecutable):
- bytecode/UnlinkedCodeBlock.h:
(JSC::UnlinkedFunctionExecutable::functionMode):
- 12:34 PM Changeset in webkit [181827] by
-
- 1 edit2 adds in trunk/LayoutTests
Import the blink test fast/css/currentcolor-strict-mode.html
https://bugs.webkit.org/show_bug.cgi?id=142838
Patch by Benjamin Poulain <bpoulain@apple.com> on 2015-03-21
Reviewed by Darin Adler.
- fast/css/currentcolor-strict-mode-expected.html: Added.
- fast/css/currentcolor-strict-mode.html: Added.
- 8:48 AM Changeset in webkit [181826] by
-
- 2 edits in trunk/Source/WebKit2
[EFL][GTK] Make WebKit2's generate-forwarding-headers.pl less verbose
https://bugs.webkit.org/show_bug.cgi?id=142908
Reviewed by Žan Doberšek.
- Scripts/generate-forwarding-headers.pl:
(createForwardingHeadersForFramework):
- 12:37 AM Changeset in webkit [181825] by
-
- 37 edits5 deletes in trunk
Unreviewed, rolling out r181824.
https://bugs.webkit.org/show_bug.cgi?id=142935
Broke Windows build (Requested by ap on #webkit).
Reverted changeset:
"Remove the prefix for CSS Transforms"
https://bugs.webkit.org/show_bug.cgi?id=142927
http://trac.webkit.org/changeset/181824
Mar 20, 2015:
- 11:54 PM Changeset in webkit [181824] by
-
- 37 edits1 copy4 adds in trunk
Remove the prefix for CSS Transforms
https://bugs.webkit.org/show_bug.cgi?id=142927
<rdar://problem/11568151>
Reviewed by Simon Fraser.
Source/WebCore:
Support the bare form of:
- transform
- transform-style
- transform-origin
- perspective
- perspective-origin
For transform-style, we also still differentiate the prefixed property
because we may want to use that as an indicator of legacy content.
This is mostly a find and replace, but with some related naming
changes. The DOM interface for WebKitCSSTransformValue remains
unchanged.
Tests: transforms/2d/transform-2d-prefixed.html
transforms/3d/general/prefixed-3dtransform-values.html
- bindings/js/JSCSSStyleDeclarationCustom.cpp:
(WebCore::propertyChangeMayRepaintNonDescendants):
- css/CSSComputedStyleDeclaration.cpp:
(WebCore::isLayoutDependent):
(WebCore::ComputedStyleExtractor::propertyValue):
- css/CSSParser.cpp:
(WebCore::isValidKeywordPropertyAndValue):
(WebCore::isKeywordPropertyID):
(WebCore::parseTranslateTransformValue):
(WebCore::CSSParser::parseValue):
(WebCore::CSSParser::parseTransformOrigin):
(WebCore::CSSParser::parsePerspectiveOrigin):
- css/CSSPropertyNames.in:
- css/StyleProperties.cpp:
(WebCore::StyleProperties::getPropertyValue):
(WebCore::StyleProperties::asText):
- css/WebKitCSSMatrix.cpp:
(WebCore::WebKitCSSMatrix::setMatrixValue):
- html/track/VTTCue.cpp:
(WebCore::VTTCueBox::applyCSSProperties):
- page/animation/AnimationController.cpp:
(WebCore::AnimationControllerPrivate::computeExtentOfAnimation):
- page/animation/CSSPropertyAnimation.cpp:
(WebCore::PropertyWrapperAcceleratedTransform::PropertyWrapperAcceleratedTransform):
(WebCore::CSSPropertyAnimationWrapperMap::CSSPropertyAnimationWrapperMap):
- page/animation/CompositeAnimation.cpp:
(WebCore::CompositeAnimation::computeExtentOfTransformAnimation):
- page/animation/ImplicitAnimation.cpp:
(WebCore::ImplicitAnimation::computeExtentOfTransformAnimation):
- page/animation/KeyframeAnimation.cpp:
(WebCore::KeyframeAnimation::computeExtentOfTransformAnimation):
(WebCore::KeyframeAnimation::validateTransformFunctionList):
- platform/graphics/GraphicsLayer.cpp:
(WebCore::GraphicsLayer::validateTransformOperations):
- platform/graphics/GraphicsLayerClient.h:
- platform/graphics/ca/GraphicsLayerCA.cpp:
(WebCore::propertyIdToString):
(WebCore::GraphicsLayerCA::moveOrCopyAnimations):
(WebCore::GraphicsLayerCA::addAnimation):
(WebCore::GraphicsLayerCA::isRunningTransformAnimation):
(WebCore::GraphicsLayerCA::createAnimationFromKeyframes):
(WebCore::GraphicsLayerCA::createTransformAnimationsFromKeyframes):
- platform/graphics/texmap/GraphicsLayerTextureMapper.cpp:
(WebCore::GraphicsLayerTextureMapper::addAnimation):
- platform/graphics/texmap/TextureMapperAnimation.cpp:
(WebCore::TextureMapperAnimation::applyInternal):
- platform/graphics/texmap/TextureMapperLayer.cpp:
(WebCore::TextureMapperLayer::syncAnimations):
- platform/graphics/texmap/coordinated/CoordinatedGraphicsLayer.cpp:
(WebCore::CoordinatedGraphicsLayer::selfOrAncestorHasActiveTransformAnimation):
(WebCore::CoordinatedGraphicsLayer::selfOrAncestorHaveNonAffineTransforms):
(WebCore::CoordinatedGraphicsLayer::addAnimation):
- rendering/RenderLayerBacking.cpp:
(WebCore::RenderLayerBacking::updateGeometry):
(WebCore::RenderLayerBacking::startAnimation):
(WebCore::RenderLayerBacking::startTransition):
(WebCore::RenderLayerBacking::graphicsLayerToCSSProperty):
(WebCore::RenderLayerBacking::cssToGraphicsLayerProperty):
- rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::requiresCompositingForAnimation):
(WebCore::RenderLayerCompositor::isRunningTransformAnimation):
- svg/SVGElement.cpp:
(WebCore::populateAttributeNameToCSSPropertyIDMap):
Source/WebKit2:
Rename AnimatedPropertyWebkitTransform to AnimatedPropertyTransform.
- Shared/CoordinatedGraphics/CoordinatedGraphicsArgumentCoders.cpp:
(IPC::ArgumentCoder<TextureMapperAnimation>::encode):
LayoutTests:
Convert the helper functions for transforms to use the unprefixed
form, and copy some tests to make sure the prefixed form still works.
- fast/css/getComputedStyle/computed-style-expected.txt:
- fast/css/getComputedStyle/computed-style-without-renderer-expected.txt:
- fast/css/getComputedStyle/resources/property-names.js:
- svg/css/getComputedStyle-basic-expected.txt:
- transforms/2d/computed-style-origin-expected.txt:
- transforms/2d/computed-style-origin.html:
- transforms/2d/resources/prefixed-transform-test-utils.js: Copied from LayoutTests/transforms/2d/resources/transform-test-utils.js.
(floatingPointEqual):
(compareMatrices):
(testTransforms):
- transforms/2d/resources/transform-test-utils.js:
(testTransforms):
- transforms/2d/transform-2d-prefixed-expected.txt: Added.
- transforms/2d/transform-2d-prefixed.html: Added.
- transforms/3d/general/prefixed-3dtransform-values-expected.txt: Added.
- transforms/3d/general/prefixed-3dtransform-values.html: Added.
- transitions/transition-end-event-multiple-04-expected.txt:
- transitions/transition-end-event-multiple-04.html:
- transitions/transition-end-event-transform-expected.txt:
- transitions/transition-end-event-transform.html:
- 9:20 PM Changeset in webkit [181823] by
-
- 2 edits in trunk/LayoutTests
[Win] Additional debug assertion documentation.
- platform/win/TestExpectations:
- 8:09 PM Changeset in webkit [181822] by
-
- 4 edits in branches/safari-600.5-branch/LayoutTests
Test gardening. Results are as expected based on code changes
on this branch.
- css3/blending/blend-mode-blended-element-overlapping-composited-sibling-should-have-compositing-layer-expected.txt:
- css3/blending/blend-mode-parent-of-composited-blended-has-layer-expected.txt:
- css3/blending/blend-mode-with-composited-descendant-should-have-layer-expected.txt:
- 7:02 PM Changeset in webkit [181821] by
-
- 3 edits1 add in trunk
GCTimer should know keep track of nested GC phases
https://bugs.webkit.org/show_bug.cgi?id=142675
Reviewed by Darin Adler.
Source/JavaScriptCore:
This improves the GC phase timing output in Heap.cpp by linking
phases nested inside other phases together, allowing tools
to compute how much time we're spending in various nested phases.
- heap/Heap.cpp:
Tools:
Adds a tool to aid in parsing the GC phase timing output into a
tree-like structure based on the parent-child relationships
of nested GC phases.
- Scripts/parse-gc-phase-timings: Added.
(Timing):
(Timing.init):
(Timing.unicode):
(Timing.str):
(Timing.repr):
(parse_input):
(print_timing_node):
(print_timing_tree):
(link_parents):
(main):
- 5:07 PM Changeset in webkit [181820] by
-
- 2 edits in trunk/LayoutTests
[Win] More debug assertions. Almost there!
- platform/win/TestExpectations:
- 4:38 PM Changeset in webkit [181819] by
-
- 2 edits in trunk/Source/WebInspectorUI
Web Inspector: Fast typing lags in the Styles sidebar or Console
https://bugs.webkit.org/show_bug.cgi?id=142919
Asynchronous autocomplete causes a race condition in CodeMirror,
which results in skipped characters in while typing. Completing immediately
fixes the issues.
Reviewed by Timothy Hatcher.
- UserInterface/Controllers/CodeMirrorCompletionController.js:
(WebInspector.CodeMirrorCompletionController.prototype.hideCompletions):
(WebInspector.CodeMirrorCompletionController.prototype._completeAtCurrentPosition):
(WebInspector.CodeMirrorCompletionController.prototype.):
- 4:37 PM Changeset in webkit [181818] by
-
- 7 edits in trunk/Source/JavaScriptCore
FunctionBodyNode should known where its parameters started
https://bugs.webkit.org/show_bug.cgi?id=142926
Reviewed by Ryosuke Niwa.
This will allow us to re-parse parameters instead of keeping the
parameters piece of the AST around forever.
I also took the opportunity to initialize most FunctionBodyNode data
members at construction time, to help clarify that they are set right.
- parser/ASTBuilder.h:
(JSC::ASTBuilder::createFunctionExpr): No need to pass
functionKeywordStart here; we now provide it at FunctionBodyNode
creation time.
(JSC::ASTBuilder::createFunctionBody): Require everything we need at
construction time, including the start of our parameters.
(JSC::ASTBuilder::createGetterOrSetterProperty):
(JSC::ASTBuilder::createFuncDeclStatement): No need to pass
functionKeywordStart here; we now provide it at FunctionBodyNode
creation time.
(JSC::ASTBuilder::setFunctionNameStart): Deleted.
- parser/Nodes.cpp:
(JSC::FunctionBodyNode::FunctionBodyNode): Initialize everything at
construction time.
- parser/Nodes.h: Added a field for the location of our parameters.
- parser/Parser.cpp:
(JSC::Parser<LexerType>::parseFunctionBody):
(JSC::Parser<LexerType>::parseFunctionInfo):
(JSC::Parser<LexerType>::parseFunctionDeclaration):
(JSC::Parser<LexerType>::parseClass):
(JSC::Parser<LexerType>::parsePropertyMethod):
(JSC::Parser<LexerType>::parseGetterSetter):
(JSC::Parser<LexerType>::parsePrimaryExpression):
- parser/Parser.h: Refactored to match above interface changes.
- parser/SyntaxChecker.h:
(JSC::SyntaxChecker::createFunctionExpr):
(JSC::SyntaxChecker::createFunctionBody):
(JSC::SyntaxChecker::createFuncDeclStatement):
(JSC::SyntaxChecker::createGetterOrSetterProperty): Refactored to match
above interface changes.
(JSC::SyntaxChecker::setFunctionNameStart): Deleted.
- 4:26 PM Changeset in webkit [181817] by
-
- 5 edits2 adds in trunk/Source/JavaScriptCore
Observably effectful nodes in DFG IR should come last in their bytecode instruction (i.e. forExit section), except for Hint nodes
https://bugs.webkit.org/show_bug.cgi?id=142920
Reviewed by Oliver Hunt, Geoffrey Garen, and Mark Lam.
Observably effectful, n.: If we reexecute the bytecode instruction after this node has
executed, then something other than the bytecode instruction's specified outcome will
happen.
We almost never had observably effectful nodes except at the end of the bytecode
instruction. The exception is a lowered transitioning PutById:
PutStructure(@o, S1 -> S2)
PutByOffset(@o, @o, @v)
The PutStructure is observably effectful: if you try to reexecute the bytecode after
doing the PutStructure, then we'll most likely crash. The generic PutById handling means
first checking what the old structure of the object is; but if we reexecute, the old
structure will seem to be the new structure. But the property ensured by the new
structure hasn't been stored yet, so any attempt to load it or scan it will crash.
Intriguingly, however, none of the other operations involved in the PutById are
observably effectful. Consider this example:
PutByOffset(@o, @o, @v)
PutStructure(@o, S1 -> S2)
Note that the PutStructure node doesn't reallocate property storage; see further below
for an example that does that. Because no property storage is happening, we know that we
already had room for the new property. This means that the PutByOffset is no observable
until the PutStructure executes and "reveals" the property. Hence, PutByOffset is not
observably effectful.
Now consider this:
b: AllocatePropertyStorage(@o)
PutByOffset(@b, @o, @v)
PutStructure(@o, S1 -> S2)
Surprisingly, this is also safe, because the AllocatePropertyStorage is not observably
effectful. It *does* reallocate the property storage and the new property storage pointer
is stored into the object. But until the PutStructure occurs, the world will just think
that the reallocation didn't happen, in the sense that we'll think that the property
storage is using less memory than what we just allocated. That's harmless.
The AllocatePropertyStorage is safe in other ways, too. Even if we GC'd after the
AllocatePropertyStorage but before the PutByOffset (or before the PutStructure),
everything could be expected to be fine, so long as all of @o, @v and @b are on the
stack. If they are all on the stack, then the GC will leave the property storage alone
(so the extra memory we just allocated would be safe). The GC will not scan the part of
the property storage that contains @v, but that's fine, so long as @v is on the stack.
The better long-term solution is probably bug 142921.
But for now, this:
- Fixes an object materialization bug, exemplified by the two tests, that previously crashed 100% of the time with FTL enabled and concurrent JIT disabled.
- Allows us to remove the workaround introduced in r174856.
- dfg/DFGByteCodeParser.cpp:
(JSC::DFG::ByteCodeParser::handlePutById):
- dfg/DFGConstantFoldingPhase.cpp:
(JSC::DFG::ConstantFoldingPhase::emitPutByOffset):
- dfg/DFGFixupPhase.cpp:
(JSC::DFG::FixupPhase::insertCheck):
(JSC::DFG::FixupPhase::indexOfNode): Deleted.
(JSC::DFG::FixupPhase::indexOfFirstNodeOfExitOrigin): Deleted.
- dfg/DFGInsertionSet.h:
(JSC::DFG::InsertionSet::insertOutOfOrder): Deleted.
(JSC::DFG::InsertionSet::insertOutOfOrderNode): Deleted.
- tests/stress/materialize-past-butterfly-allocation.js: Added.
(bar):
(foo0):
(foo1):
(foo2):
(foo3):
(foo4):
- tests/stress/materialize-past-put-structure.js: Added.
(foo):
- 4:09 PM Changeset in webkit [181816] by
-
- 3 edits in trunk/Source/WebKit2
[WK2] NetworkCache retrievals sometimes fail on browser startup
https://bugs.webkit.org/show_bug.cgi?id=142925
<rdar://problem/20245368>
Reviewed by Antti Koivisto.
NetworkCache retrievals sometimes fail on browser startup for resources
that are actually cached. The reason is that we are using a bloom filter
for performance reasons to avoid unnecessary disk I/O and this bloom
filter is populated on start up in a background thread by traversing the
cache files on disk. However, when restoring the tabs on start-up we
sometimes query this bloom filter before it is completely populated and
we thus fail to retrieve cached entries because we think they don't
exist and don't check the disk.
This patch adds an "isPopulatingContentsFilter" flag that is turned ON
on start up while we are populating the bloon filter. We then bypass
the bloom filter and send queries directly to disk on start up if this
flag is ON.
- NetworkProcess/cache/NetworkCacheStorage.cpp:
(WebKit::NetworkCache::Storage::initialize):
(WebKit::NetworkCache::Storage::retrieve):
(WebKit::NetworkCache::Storage::dispatchPendingWriteOperations):
(WebKit::NetworkCache::Storage::dispatchHeaderWriteOperation):
- NetworkProcess/cache/NetworkCacheStorage.h:
(WebKit::NetworkCache::Storage::cacheMayContain):
- 3:45 PM Changeset in webkit [181815] by
-
- 6 edits in trunk/Source/WebKit2
[WK2] Allow stale content when restoring the browser's session state
https://bugs.webkit.org/show_bug.cgi?id=142916
<rdar://problem/20243493>
Reviewed by Darin Adler.
Allow stale content when restoring the browser's session state
(restoring all tabs from previous session), e.g.
- Via History > Reopen All windows from previous session on Safari
- Happens on startup for MobileSafari
I have verified that using "Open in tabs" on a bookmarks folder
still does fresh loads (rdar://problem/8131355) as it is not using
the restoreFromSessionState() code path.
- UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::reattachToWebProcessForReload):
(WebKit::WebPageProxy::reattachToWebProcessWithItem):
(WebKit::WebPageProxy::goForward):
(WebKit::WebPageProxy::goBack):
(WebKit::WebPageProxy::goToBackForwardItem):
(WebKit::WebPageProxy::restoreFromSessionState):
- UIProcess/WebPageProxy.h:
- WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::goToBackForwardItem):
- WebProcess/WebPage/WebPage.h:
- WebProcess/WebPage/WebPage.messages.in:
- 2:35 PM Changeset in webkit [181814] by
-
- 17 edits in trunk/Source
REGRESSION (r179429): Potential Use after free in JavaScriptCore`WTF::StringImpl::ref + 83
https://bugs.webkit.org/show_bug.cgi?id=142410
Reviewed by Geoffrey Garen.
Source/JavaScriptCore:
Before this patch, added function JSValue::toPropertyKey returns PropertyName.
Since PropertyName doesn't have AtomicStringImpl ownership,
if Identifier is implicitly converted to PropertyName and Identifier is destructed,
PropertyName may refer freed AtomicStringImpl*.
This patch changes the result type of JSValue::toPropertyName from PropertyName to Identifier,
to keep AtomicStringImpl* ownership after the toPropertyName call is done.
And receive the result value as Identifier type to keep ownership in the caller side.
To catch the result of toPropertyKey as is, we catch the result of toPropertyName as auto.
However, now we don't need to have both Identifier and PropertyName.
So we'll merge PropertyName to Identifier in the subsequent patch.
- dfg/DFGOperations.cpp:
(JSC::DFG::operationPutByValInternal):
- jit/JITOperations.cpp:
(JSC::getByVal):
- llint/LLIntSlowPaths.cpp:
(JSC::LLInt::getByVal):
(JSC::LLInt::LLINT_SLOW_PATH_DECL):
- runtime/CommonSlowPaths.cpp:
(JSC::SLOW_PATH_DECL):
- runtime/CommonSlowPaths.h:
(JSC::CommonSlowPaths::opIn):
- runtime/JSCJSValue.h:
- runtime/JSCJSValueInlines.h:
(JSC::JSValue::toPropertyKey):
- runtime/ObjectConstructor.cpp:
(JSC::objectConstructorGetOwnPropertyDescriptor):
(JSC::objectConstructorDefineProperty):
- runtime/ObjectPrototype.cpp:
(JSC::objectProtoFuncPropertyIsEnumerable):
Source/WebCore:
The same issues are found in the existing code; PropertyName does not have ownership.
This patch rewrite the point that should have ownership to Identifier.
- bindings/js/JSDOMWindowCustom.cpp:
(WebCore::JSDOMWindow::getOwnPropertySlotByIndex):
(WebCore::JSDOMWindow::putByIndex):
- bindings/js/ReadableStreamJSSource.cpp:
(WebCore::getInternalSlotFromObject):
- bindings/scripts/CodeGeneratorJS.pm:
(GenerateImplementation):
- bindings/scripts/test/JS/JSTestCustomNamedGetter.cpp:
(WebCore::JSTestCustomNamedGetter::getOwnPropertySlotByIndex):
- bindings/scripts/test/JS/JSTestEventTarget.cpp:
(WebCore::JSTestEventTarget::getOwnPropertySlotByIndex):
- bindings/scripts/test/JS/JSTestInterface.cpp:
(WebCore::JSTestInterface::putByIndex):
- 2:32 PM Changeset in webkit [181813] by
-
- 3 edits in trunk/Source/WebCore
Log the slow-scrolling reasons (if any) on first commit of a ScrollingTreeFrameScrollingNodeMac
https://bugs.webkit.org/show_bug.cgi?id=142917
Reviewed by Beth Dakin.
The scrollperf test would fail to detect transitions between slow and fast scrolling
for new pages, because there was no logging when slow-scrolling scrolling tree was
torn down, and a new fast-scrolling tree constructed.
Fix by always logging the slow-scrolling reasons in the first update of the
ScrollingTreeFrameScrollingNodeMac, even when the slow-scrolling reasons haven't
changed.
- page/scrolling/mac/ScrollingTreeFrameScrollingNodeMac.h:
- page/scrolling/mac/ScrollingTreeFrameScrollingNodeMac.mm:
(WebCore::ScrollingTreeFrameScrollingNodeMac::ScrollingTreeFrameScrollingNodeMac):
(WebCore::ScrollingTreeFrameScrollingNodeMac::updateBeforeChildren):
- 2:23 PM Changeset in webkit [181812] by
-
- 2 edits in trunk/Source/WebKit2
[Immediate Actions] Should re-enable immediate actions for iBooks
https://bugs.webkit.org/show_bug.cgi?id=142915
-and corresponding-
rdar://problem/19585689
Reviewed by Tim Horton.
- UIProcess/API/mac/WKView.mm:
(-[WKView initWithFrame:processPool:configuration:webView:]):
- 2:16 PM Changeset in webkit [181811] by
-
- 2 edits in trunk/LayoutTests
[Win] Attempting to get EWS green.
- platform/win/TestExpectations:
- 1:12 PM Changeset in webkit [181810] by
-
- 45 edits3 deletes in trunk
Function.prototype.toString should not decompile the AST
https://bugs.webkit.org/show_bug.cgi?id=142853
Reviewed by Sam Weinig.
Source/JavaScriptCore:
To recover the function parameter string, Function.prototype.toString
decompiles the function parameters from the AST. This is bad for a few
reasons:
(1) It requires us to keep pieces of the AST live forever. This is an
awkward design and a waste of memory.
(2) It doesn't match Firefox or Chrome (because it changes whitespace
and ES6 destructuring expressions).
(3) It doesn't scale to ES6 default argument parameters, which require
arbitrarily complex decompilation.
(4) It can counterfeit all the line numbers in a function (because
whitespace can include newlines).
(5) It's expensive, and we've seen cases where websites invoke
Function.prototype.toString a lot by accident.
The fix is to do what we do for the rest of the function: Just quote the
original source text.
Since this change inevitably changes some function stringification, I
took the opportunity to make our stringification match Firefox's and
Chrome's.
- API/tests/testapi.c:
(assertEqualsAsUTF8String): Be more informative when this fails.
(main): Updated to match new stringification rules.
- bytecode/UnlinkedCodeBlock.cpp:
(JSC::UnlinkedFunctionExecutable::paramString): Deleted. Yay!
- bytecode/UnlinkedCodeBlock.h:
- parser/Nodes.h:
(JSC::StatementNode::isFuncDeclNode): New helper for constructing
anonymous functions.
- parser/SourceCode.h:
(JSC::SourceCode::SourceCode): Allow zero because WebCore wants it.
- runtime/CodeCache.cpp:
(JSC::CodeCache::getFunctionExecutableFromGlobalCode): Updated for use
of function declaration over function expression.
- runtime/Executable.cpp:
(JSC::FunctionExecutable::paramString): Deleted. Yay!
- runtime/Executable.h:
(JSC::FunctionExecutable::parameterCount):
- runtime/FunctionConstructor.cpp:
(JSC::constructFunctionSkippingEvalEnabledCheck): Added a newline after
the opening brace to match Firefox and Chrome, and a space after the comma
to match Firefox and WebKit coding style. Added the function name to
the text of the function so it would look right when stringify-ing. Switched
from parentheses to braces to produce a function declaration instead of
a function expression because we are required to exclude the function's
name from its scope, and that's what a function declaration does.
- runtime/FunctionPrototype.cpp:
(JSC::functionProtoFuncToString): Removed an old workaround because the
library it worked around doesn't really exist anymore, and the behavior
doesn't match Firefox or Chrome. Use type profiling offsets instead of
function body offsets because we want to include the function name and
the parameter string, rather than stitching them in manually by
decompiling the AST.
(JSC::insertSemicolonIfNeeded): Deleted.
- tests/mozilla/js1_2/function/tostring-1.js:
- tests/mozilla/js1_5/Scope/regress-185485.js:
(with.g): Updated these test results for formatting changes.
Source/WebCore:
- bindings/js/JSLazyEventListener.cpp:
(WebCore::JSLazyEventListener::initializeJSFunction): Adjust the line
number of attribute event listeners to account for the leading newline
now added by JavaScriptCore.
This solution is not perfect, but there are a lot of pre-existing problems
with line and column reporting for attribute event listeners, and this
preserves existing behavior with reasonable reliability.
LayoutTests:
Updated test results to match new rules for Function.prototype.toString.
- fast/dom/TreeWalker/acceptNode-filter-expected.txt: Removed a space
because it was not in the original source.
- fast/events/window-onerror2-expected.txt: Column number changed because
the event listener body starts on its own line now. This was a bit wrong
before and is still a bit wrong now in a different way.
- fast/profiler/dead-time-expected.txt:
- fast/profiler/inline-event-handler-expected.txt:
- fast/profiler/stop-profiling-after-setTimeout-expected.txt: Line number
changed because WebCore shifts line nubmers on attribute event listeners
by one.
- js/class-syntax-default-constructor-expected.txt: Constructor name
is not present now because it is not present in the source text. This
test failed before and it still fails now in a slightly different way.
- js/destructuring-assignment-expected.txt: Destructuring arguments now
match their source text faithfully.
- js/dfg-redundant-load-of-captured-variable-proven-constant-expected.txt:
Removed a space because it was not present in the original source text.
- js/dfg-resolve-global-specific-dictionary-expected.txt: Ditto.
- js/function-toString-semicolon-insertion-expected.txt: Removed.
- js/script-tests/function-toString-semicolon-insertion.js: Removed.
- js/function-toString-semicolon-insertion.html: Removed. This test checked
for a work-around that I have removed.
- js/object-literal-computed-methods-expected.txt:
- js/object-literal-methods-expected.txt: These tests fail because object
literal methods do not register their function names appropriately. This
was a pre-existing failure that is now more explicit.
- js/dom/JSON-parse-expected.txt:
- js/dom/JSON-stringify-expected.txt: Whitespace removed because it was
not present in the original.
- js/dom/dfg-strcat-over-objects-then-exit-on-it-expected.txt: Ditto.
- js/dom/function-prototype-expected.txt:
- js/dom/function-prototype.html: Ditto.
- js/dom/parse-error-external-script-in-new-Function-expected.txt: Line
changed by one due to new extra newline.
- js/dom/script-start-end-locations-expected.txt: Lines and columns
changed due to new extra newline.
- js/dom/toString-and-valueOf-override-expected.txt: Whitespace removed
because it was not present in the original.
- js/dom/script-tests/dfg-strcat-over-objects-then-exit-on-it.js: Ditto.
- js/kde/lval-exceptions-expected.txt: Ditto.
- js/script-tests/dfg-redundant-load-of-captured-variable-proven-constant.js: Ditto.
- js/script-tests/dfg-resolve-global-specific-dictionary.js: Ditto.
- platform/mac/http/tests/media/media-source/mediasource-sourcebuffer-mode-expected.txt: Ditto.
- storage/domstorage/localstorage/string-conversion-expected.txt: Ditto.
- storage/domstorage/sessionstorage/string-conversion-expected.txt: Ditto.
- userscripts/window-onerror-for-isolated-world-1-expected.txt:
- userscripts/window-onerror-for-isolated-world-2-expected.txt: Line numbers
changed because of new anonymous function formatting. These line numbers
were wrong before and they are still wrong now.
- 12:14 PM Changeset in webkit [181809] by
-
- 2 edits in trunk/LayoutTests
[Win] Add some skips to support EWS use.
- platform/win/TestExpectations:
- 11:18 AM Changeset in webkit [181808] by
-
- 9 edits1 add in trunk/Source
Progress towards CMake on Mac.
https://bugs.webkit.org/show_bug.cgi?id=136003
Reviewed by Chris Dumez.
Source/WebCore:
- PlatformMac.cmake:
Added some directories to create forwarding headers from.
Generate ObjC bindings.
Source/WebKit:
- CMakeLists.txt:
- PlatformMac.cmake: Added.
Source/WebKit2:
- PlatformMac.cmake:
Add include directories and forwarding headers.
- Shared/API/APIObject.h:
(API::ObjectImpl::operator new):
- UIProcess/DrawingAreaProxy.h:
(WebKit::DrawingAreaProxy::intrinsicContentSizeDidChange):
- UIProcess/mac/TiledCoreAnimationDrawingAreaProxy.h:
Fix some unused parameter warnings.
- 11:10 AM Changeset in webkit [181807] by
-
- 5 edits in trunk
SyntaxChecker assertion is trapped with computed property name and getter
https://bugs.webkit.org/show_bug.cgi?id=142863
Patch by Joseph Pecoraro <Joseph Pecoraro> on 2015-03-20
Reviewed by Ryosuke Niwa.
Source/JavaScriptCore:
- parser/SyntaxChecker.h:
(JSC::SyntaxChecker::getName):
Remove invalid assert. Computed properties will not have a name
and the calling code is checking for null expecting it. The
AST path (non-CheckingPath) already does this without the assert
so it is well tested.
LayoutTests:
- js/basic-computed-property-name-expected.txt:
- js/script-tests/basic-computed-property-name.js:
(a.string_appeared_here.runTest):
Add tests that wrap the expression in a function, this
will test the syntax checker.
- 11:08 AM Changeset in webkit [181806] by
-
- 9 edits2 adds in trunk/Source/JavaScriptCore
JSCallbackObject<JSGlobalObject> should not destroy its JSCallbackObjectData before all its finalizers have been called.
<https://webkit.org/b/142846>
Reviewed by Geoffrey Garen.
Currently, JSCallbackObject<JSGlobalObject> registers weak finalizers via 2 mechanisms:
- JSCallbackObject<Parent>::init() registers a weak finalizer for all JSClassRef that a JSCallbackObject references.
- JSCallbackObject<JSGlobalObject>::create() registers a finalizer via vm.heap.addFinalizer() which destroys the JSCallbackObject.
The first finalizer is implemented as a virtual function of a JSCallbackObjectData
instance that will be destructed if the 2nd finalizer is called. Hence, if the
2nd finalizer if called first, the later invocation of the 1st finalizer will
result in a crash.
This patch fixes the issue by eliminating the finalizer registration in init().
Instead, we'll have the JSCallbackObject destructor call all the JSClassRef finalizers
if needed. This ensures that these finalizers are called before the JSCallbackObject
is destructor.
Also added assertions to a few Heap functions because JSCell::classInfo() expects
all objects that are allocated from MarkedBlock::Normal blocks to be derived from
JSDestructibleObject. These assertions will help us catch violations of this
expectation earlier.
- API/JSCallbackObject.cpp:
(JSC::JSCallbackObjectData::finalize): Deleted.
- API/JSCallbackObject.h:
(JSC::JSCallbackObjectData::~JSCallbackObjectData):
- API/JSCallbackObjectFunctions.h:
(JSC::JSCallbackObject<Parent>::~JSCallbackObject):
(JSC::JSCallbackObject<Parent>::init):
- API/tests/GlobalContextWithFinalizerTest.cpp: Added.
(finalize):
(testGlobalContextWithFinalizer):
- API/tests/GlobalContextWithFinalizerTest.h: Added.
- API/tests/testapi.c:
(main):
- JavaScriptCore.vcxproj/testapi/testapi.vcxproj:
- JavaScriptCore.vcxproj/testapi/testapi.vcxproj.filters:
- JavaScriptCore.xcodeproj/project.pbxproj:
- heap/HeapInlines.h:
(JSC::Heap::allocateObjectOfType):
(JSC::Heap::subspaceForObjectOfType):
(JSC::Heap::allocatorForObjectOfType):
- 10:53 AM Changeset in webkit [181805] by
-
- 3 edits in trunk/Source/WebCore
[WinCairo] Crash when leaving page while video is playing.
https://bugs.webkit.org/show_bug.cgi?id=142578
Reviewed by Alex Christensen.
This can happen when there is an asynchronous event invoked from a video worker
thread, accessing the media player object after it has been deleted.
This is solved by resetting the media player pointer in the event object,
when the media player is deleted.
- platform/graphics/win/MediaPlayerPrivateMediaFoundation.cpp:
(WebCore::MediaPlayerPrivateMediaFoundation::~MediaPlayerPrivateMediaFoundation):
(WebCore::MediaPlayerPrivateMediaFoundation::addListener):
(WebCore::MediaPlayerPrivateMediaFoundation::removeListener):
(WebCore::MediaPlayerPrivateMediaFoundation::notifyDeleted):
(WebCore::MediaPlayerPrivateMediaFoundation::AsyncCallback::AsyncCallback):
(WebCore::MediaPlayerPrivateMediaFoundation::AsyncCallback::~AsyncCallback):
(WebCore::MediaPlayerPrivateMediaFoundation::AsyncCallback::Invoke):
(WebCore::MediaPlayerPrivateMediaFoundation::AsyncCallback::onMediaPlayerDeleted):
- platform/graphics/win/MediaPlayerPrivateMediaFoundation.h:
(WebCore::MediaPlayerPrivateMediaFoundation::MediaPlayerListener::MediaPlayerListener):
(WebCore::MediaPlayerPrivateMediaFoundation::MediaPlayerListener::~MediaPlayerListener):
(WebCore::MediaPlayerPrivateMediaFoundation::MediaPlayerListener::onMediaPlayerDeleted):
- 10:18 AM Changeset in webkit [181804] by
-
- 2 edits in trunk/Source/WebCore
Regression(r181782): Caused WebKit2.RestoreSessionStateContainingFormData API test to crash
https://bugs.webkit.org/show_bug.cgi?id=142911
Reviewed by Antti Koivisto.
Drop the isBackForwardLoadType(m_loadType) assertion in
FrameLoader::retryAfterFailedCacheOnlyMainResourceLoad(). This assertion
used to hold true when goToBackForwardItem() was only called with
history navigation load types. However, after r181782, we use a Standard
load in the cases where we do not wish to display stale content (e.g.
restoring session state).
The assertion comment says "We only use cache-only loads to avoid
resubmitting forms". However, we then call loadDifferentDocumentItem()
which already takes care of using the ReturnCacheDataElseLoad cache
policy as the HistoryItem has form data and we pass
MayNotAttemptCacheOnlyLoadForFormSubmissionItem as parameter.
No new tests, covered by WebKit2.RestoreSessionStateContainingFormData
API test.
- loader/FrameLoader.cpp:
(WebCore::FrameLoader::retryAfterFailedCacheOnlyMainResourceLoad):
- 10:07 AM Changeset in webkit [181803] by
-
- 3 edits in trunk/LayoutTests
platform/mac/fast/loader/file-url-mimetypes-3.html fails when MS Office is installed
https://bugs.webkit.org/show_bug.cgi?id=142884
Reviewed by Tim Horton.
Disable a test for .xls until rdar://problem/20232251 is fixed. It is not particularly
relevant to a web browser which MIME type local Excel files get anyway.
- platform/mac/fast/loader/file-url-mimetypes-3-expected.txt:
- platform/mac/fast/loader/file-url-mimetypes-3.html:
- 9:47 AM Changeset in webkit [181802] by
-
- 1 edit1 add9 deletes in trunk/LayoutTests
[Mac] svg/text/non-bmp-positioning-lists.svg fails when MS Office is installed
https://bugs.webkit.org/show_bug.cgi?id=142861
Reviewed by Tim Horton.
Converted into a reftest. It would be nice to make tests not depend on installed
fonts in general, but that doesn't seem very feasible (rdar://problem/20000350).
- platform/efl/svg/text/non-bmp-positioning-lists-expected.png: Removed.
- platform/efl/svg/text/non-bmp-positioning-lists-expected.txt: Removed.
- platform/gtk/svg/text/non-bmp-positioning-lists-expected.png: Removed.
- platform/gtk/svg/text/non-bmp-positioning-lists-expected.txt: Removed.
- platform/ios-sim-deprecated/svg/text/non-bmp-positioning-lists-expected.txt: Removed.
- platform/ios-simulator/svg/text/non-bmp-positioning-lists-expected.txt: Removed.
- platform/mac/svg/text/non-bmp-positioning-lists-expected.png: Removed.
- platform/mac/svg/text/non-bmp-positioning-lists-expected.txt: Removed.
- platform/win/svg/text/non-bmp-positioning-lists-expected.txt: Removed.
- svg/text/non-bmp-positioning-lists-expected.svg: Added.
- 8:57 AM Changeset in webkit [181801] by
-
- 2 edits in trunk/Tools
[Win] Temporarily turn of EWS Windows tests while I debug the test failure.
- Scripts/webkitpy/common/config/ews.json:
- 4:26 AM Changeset in webkit [181800] by
-
- 3 edits in trunk/Source/WebKit2
[CMake] Remove duplicated cmake function calls in WebKit2/CMakeLists.txt
https://bugs.webkit.org/show_bug.cgi?id=142901
Reviewed by Csaba Osztrogonác.
add_depencencies(), install() can support multiple targets or location.
We don't need to call it twice.
- CMakeLists.txt: Additionaly move -DBUILDING_WEBKIT to PlatformGTK.cmake,
because it is only used by GTK port.
- 4:06 AM Changeset in webkit [181799] by
-
- 2 edits in releases/WebKitGTK/webkit-2.8/Source/WebKit2
Merge r181798 - [GTK] Disable accelerated compositing on Wayland
https://bugs.webkit.org/show_bug.cgi?id=142877
Reviewed by Carlos Garcia Campos.
- UIProcess/API/gtk/WebKitWebViewBase.cpp:
(webkitWebViewBaseCreateWebPage): As was done in the past, we should disable
accelerated compositing on Wayland until the proper support for it lands.
- 3:38 AM Changeset in webkit [181798] by
-
- 2 edits in trunk/Source/WebKit2
[GTK] Disable accelerated compositing on Wayland
https://bugs.webkit.org/show_bug.cgi?id=142877
Reviewed by Carlos Garcia Campos.
- UIProcess/API/gtk/WebKitWebViewBase.cpp:
(webkitWebViewBaseCreateWebPage): As was done in the past, we should disable
accelerated compositing on Wayland until the proper support for it lands.
- 3:09 AM Changeset in webkit [181797] by
-
- 2 edits in trunk/LayoutTests
Unreviewed EFL gardening. Update attachment-foo tests to failure.
https://bugs.webkit.org/show_bug.cgi?id=142653
Patch by Hunseop Jeong <Hunseop Jeong> on 2015-03-20
- platform/efl/TestExpectations:
- 2:40 AM Changeset in webkit [181796] by
-
- 2 edits in trunk/LayoutTests
Unreviewed EFL gardening. Remove duplicate or ambiguous entry lines.
https://bugs.webkit.org/show_bug.cgi?id=142775
Patch by Hunseop Jeong <Hunseop Jeong> on 2015-03-20
- platform/efl/TestExpectations:
- 2:36 AM Changeset in webkit [181795] by
-
- 3 edits in trunk/Source/WebCore
[Content Filtering] Adopt NEFilterSource unblock handling
https://bugs.webkit.org/show_bug.cgi?id=142904
Reviewed by Andreas Kling.
- platform/cocoa/NetworkExtensionContentFilter.mm:
(WebCore::NetworkExtensionContentFilter::unblockHandler):
Returned an unblock handler that calls -[NEFilterSource remediateWithDecisionHandler:].
- platform/cocoa/ParentalControlsContentFilter.mm:
(WebCore::ParentalControlsContentFilter::unblockHandler): Used ASCIILiteral.
- 2:26 AM Changeset in webkit [181794] by
-
- 2 edits in releases/WebKitGTK/webkit-2.8
Merge r181793 - [GTK] Search for the Wayland dependency when enabling Wayland target
https://bugs.webkit.org/show_bug.cgi?id=142876
Reviewed by Carlos Garcia Campos.
- Source/cmake/OptionsGTK.cmake: The Wayland dependency isn't a public
requirement of either the GTK+ or GDK pkg-config files, so we have to
search for it ourselves when WebKitGTK+ has been configured to support
the Wayland windowing target.
- 2:23 AM Changeset in webkit [181793] by
-
- 2 edits in trunk
[GTK] Search for the Wayland dependency when enabling Wayland target
https://bugs.webkit.org/show_bug.cgi?id=142876
Reviewed by Carlos Garcia Campos.
- Source/cmake/OptionsGTK.cmake: The Wayland dependency isn't a public
requirement of either the GTK+ or GDK pkg-config files, so we have to
search for it ourselves when WebKitGTK+ has been configured to support
the Wayland windowing target.
- 2:00 AM Changeset in webkit [181792] by
-
- 2 edits in trunk/Tools
[GTK][jhbuild] Fix flakey failures due to unstable googlecode repository
https://bugs.webkit.org/show_bug.cgi?id=142867
Rubber-stamped by Csaba Osztrogon.
- gtk/jhbuild.modules: Use Ossy's libusrsctp Github fork. Also use
a fixed revision for the openwebrtc-gst-plugins.
- 1:42 AM Changeset in webkit [181791] by
-
- 30 edits4 adds in trunk
[Content Filtering] Add tests for unblock requests
https://bugs.webkit.org/show_bug.cgi?id=142900
Reviewed by Andreas Kling.
Source/WebCore:
Currently the iOS Parental Controls content filter has a mechanism for requesting that a page be unblocked.
WebKit implements this by listening for navigations originating from the filter's error page to a special URL,
and requesting the page be unblocked via platform API, which might cause UI to be displayed. If the unblock is
successful then we schedule a reload of the frame in order to display the unblocked document.
NetworkExtension also supports unblock requests, so in preparation for adopting its API, this patch allows
content filters to specify their own unblock request method, teaches MockContentFilter to provide such a method,
and writes tests to cover both allowed and denied unblock requests.
The content filter that blocks a load creates a ContentFilterUnblockHandler, passing it a lambda that is executed
when a navigation matches the filter's special unblock URL. Filters can also specify that a script be executed in
the context of its error page if the unblock is denied.
All platform content filters can handle unblock requests like this with the exception of iOS Parental Controls in WebKit2.
Since UI can be displayed by the system in this case, the request must be made from within the UI process. Therefore the
existing method is retained of serializing a WebFilterEvaluator and intercepting navigation policy calls in the UI process.
Tests: contentfiltering/allow-after-unblock-request.html
contentfiltering/block-after-unblock-request.html
- bindings/js/JSMockContentFilterSettingsCustom.cpp:
(WebCore::JSMockContentFilterSettings::decisionPoint): Added some using statements for clarity.
(WebCore::JSMockContentFilterSettings::setDecisionPoint): Ditto.
(WebCore::toJSValue): Returns a JSValue from a Decision.
(WebCore::toDecision): Returns a Decision from a JSValue.
(WebCore::JSMockContentFilterSettings::decision): Used toJSValue.
(WebCore::JSMockContentFilterSettings::setDecision): Used toDecision.
(WebCore::JSMockContentFilterSettings::unblockRequestDecision): Used toJSValue.
(WebCore::JSMockContentFilterSettings::setUnblockRequestDecision): Used toDecision.
- loader/ContentFilter.cpp:
(WebCore::ContentFilter::createIfNeeded): Passed a reference to the owning DocumentLoader.
(WebCore::ContentFilter::ContentFilter): Ditto.
(WebCore::ContentFilter::unblockHandler): If the unblockHandler requests that a script be executed when an
unblock request is denied, create a wrapper unblockHandler that executes that script in m_documentLoader's frame.
- loader/ContentFilter.h:
- loader/DocumentLoader.cpp:
(WebCore::DocumentLoader::responseReceived): Passed this to ContentFilter::createIfNeeded.
- loader/FrameLoader.cpp:
(WebCore::FrameLoader::prepareForLoadStart): Called PolicyChecker::prepareForLoadStart.
- loader/PolicyChecker.cpp:
(WebCore::PolicyChecker::prepareForLoadStart): Reset m_contentFilterUnblockHandler.
(WebCore::PolicyChecker::checkNavigationPolicy): Moved logic to here from WebKit1's WebFrameLoaderClient.
Placing it here allows it to be shared between WebKit1 and WebKit2 (when the unblock handler does not need to
be called in the UI process).
- loader/PolicyChecker.h:
(WebCore::PolicyChecker::setContentFilterUnblockHandler): Added.
- page/Frame.h: Made Frame ThreadSafeRefCounted, since RefPtr<Frames> are captured in lambdas that can be
copied by background threads managed by the underlying platform.
- platform/ContentFilterUnblockHandler.h:
(WebCore::ContentFilterUnblockHandler::unblockURLScheme): Returned the Apple content filter scheme.
(WebCore::ContentFilterUnblockHandler::unblockURLHost): Returned the unblock URL host.
(WebCore::ContentFilterUnblockHandler::clear): Deleted.
- platform/PlatformContentFilter.h:
(WebCore::PlatformContentFilter::unblockRequestDeniedScript): Returned the unblock request denied script.
- platform/cocoa/ContentFilterUnblockHandlerCocoa.mm:
(WebCore::ContentFilterUnblockHandler::ContentFilterUnblockHandler): Added a constructor that takes an
unblockURLHost and a UnblockRequesterFunction. Added an alternate constructor for iOS Parental Controls on
WebKit2 that takes an unblockURLHost and a WebFilterEvaluator.
(WebCore::ContentFilterUnblockHandler::needsUIProcess): Returned true if m_webFilterEvaluator is non-null.
(WebCore::ContentFilterUnblockHandler::encode): Encoded m_unblockURLHost in addition to m_webFilterEvaluator.
(WebCore::ContentFilterUnblockHandler::decode): Decoded m_unblockURLHost in addition to m_webFilterEvaluator.
(WebCore::ContentFilterUnblockHandler::canHandleRequest): Returned true if there is a either a m_unblockRequester
or a m_webFilterEvaluator and the request's host and scheme match those of the unblock request URL.
(WebCore::dispatchToMainThread): Added a helper to dispatch a block to the main thread. Then if the web thread
is enabled on iOS, dispatch it there.
(WebCore::ContentFilterUnblockHandler::requestUnblockAsync): Renamed from handleUnblockRequestAndDispatchIfSuccessful.
Requested an unblock using either m_unblockRequester or m_webFilterEvaluator, then called decisionHandler with the response.
(WebCore::scheme): Moved to ContentFilterUnblockHandler::unblockURLScheme.
(WebCore::ContentFilterUnblockHandler::handleUnblockRequestAndDispatchIfSuccessful): Renamed to requestUnblockAsync.
- platform/cocoa/ParentalControlsContentFilter.mm:
(WebCore::ParentalControlsContentFilter::unblockHandler): Returned an unblock handler using the WebFilterEvaluator constructor.
- testing/MockContentFilter.cpp: Added using statments for clarity.
(WebCore::settings): Added a helper to get MockContentFilterSettings::singleton().
(WebCore::MockContentFilter::canHandleResponse): Used the helper.
(WebCore::MockContentFilter::MockContentFilter): Took advantage of the using statements.
(WebCore::MockContentFilter::addData): Ditto.
(WebCore::MockContentFilter::finishedAddingData): Ditto.
(WebCore::MockContentFilter::unblockHandler): Returned a ContentFilterUnblockHandler that checks settings() for its decision.
(WebCore::MockContentFilter::unblockRequestDeniedScript): Returned the script to execute in MockContentFilter's
error page when an unblock request is denied.
(WebCore::MockContentFilter::maybeDetermineStatus): Took advantage of settings() and using statements.
- testing/MockContentFilterSettings.cpp:
(WebCore::MockContentFilterSettings::unblockRequestURL): Constructed a static unblock URL and returned it.
- testing/MockContentFilterSettings.h:
(WebCore::MockContentFilterSettings::unblockURLHost): Returned the filter's unblock URL host.
(WebCore::MockContentFilterSettings::unblockRequestDecision): Returns the decision to make for an unblock request.
(WebCore::MockContentFilterSettings::setUnblockRequestDecision): Sets the decision to make for an unblock request.
- testing/MockContentFilterSettings.idl: Added the unblockRequestDecision and unblockRequestURL attributes.
Source/WebKit/mac:
- WebCoreSupport/WebFrameLoaderClient.mm:
(WebFrameLoaderClient::dispatchDidStartProvisionalLoad): This now happens in PolicyChecker.
(WebFrameLoaderClient::dispatchDecidePolicyForNavigationAction): Ditto.
- WebView/WebFrame.mm:
(-[WebFrame _contentFilterDidHandleNavigationAction:]): Deleted.
- WebView/WebFrameInternal.h: Removed contentFilterUnblockHandler from WebFramePrivate.
Source/WebKit2:
- UIProcess/Cocoa/WebPageProxyCocoa.mm:
(WebKit::WebPageProxy::contentFilterDidBlockLoadForFrame): Called WebFrameProxy::contentFilterDidBlockLoad.
- UIProcess/WebFrameProxy.cpp:
(WebKit::WebFrameProxy::didStartProvisionalLoad): Assigned a default-constructed ContentFilterUnblockHandler instead of calling clear().
(WebKit::WebFrameProxy::didHandleContentFilterUnblockNavigation): Renamed from contentFilterDidHandleNavigationAction.
Updated to use ContentFilterUnblockHandler's new API.
(WebKit::WebFrameProxy::contentFilterDidHandleNavigationAction): Deleted.
- UIProcess/WebFrameProxy.h:
(WebKit::WebFrameProxy::contentFilterDidBlockLoad): Renamed from setContentFilterUnblockHandler.
(WebKit::WebFrameProxy::setContentFilterUnblockHandler): Deleted.
- UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::decidePolicyForNavigationAction): Called WebFrameProxy::didHandleContentFilterUnblockNavigation.
- WebProcess/WebCoreSupport/WebFrameLoaderClient.cpp:
(WebKit::WebFrameLoaderClient::contentFilterDidBlockLoad): If the unblock handler needs the UI process,
send WebPageProxy::ContentFilterDidBlockLoadForFrame. Oterwise, call PolicyChecker::setContentFilterUnblockHandler.
LayoutTests:
Taught contentfiltering.js how to perform an unblock request test, and added tests for both allowed and blocked requests.
- contentfiltering/allow-after-unblock-request-expected.html: Added.
- contentfiltering/allow-after-unblock-request.html: Added.
- contentfiltering/block-after-unblock-request-expected.html: Added.
- contentfiltering/block-after-unblock-request.html: Added.
- contentfiltering/resources/contentfiltering.js:
(testContentFiltering): Added an argument specifying if the decision applies to the initial load or the unblock request.
(_doTest): When testing unblock handling, navigate the test iframe to settings.unblockRequestURL when the error page is displayed.
If the unblock is denied, the test harness will call window.unblockRequestDenied(). If the unblock is successful,
the iframe will reload, which we detect by listening for its load event.
- 1:35 AM Changeset in webkit [181790] by
-
- 2 edits in releases/WebKitGTK/webkit-2.8/Source/WebKit2
Merge r181789 - [GTK] Properly guard X11-specific code in BackingStore::createBackend()
https://bugs.webkit.org/show_bug.cgi?id=142875
Reviewed by Martin Robinson.
- UIProcess/cairo/BackingStoreCairo.cpp:
(WebKit::BackingStore::createBackend): Guard the GTK- and X11-specific
bit of code with PLATFORM(GTK) and PLATFORM(X11). Testing GDK_WINDOWING_X11
ensures that the GTK+ dependency has X11 support, but does not ensure
that WebKitGTK+ has been configured to build for X11 environments (which
is what PLATFORM(X11) ensures).
- 1:24 AM Changeset in webkit [181789] by
-
- 2 edits in trunk/Source/WebKit2
[GTK] Properly guard X11-specific code in BackingStore::createBackend()
https://bugs.webkit.org/show_bug.cgi?id=142875
Reviewed by Martin Robinson.
- UIProcess/cairo/BackingStoreCairo.cpp:
(WebKit::BackingStore::createBackend): Guard the GTK- and X11-specific
bit of code with PLATFORM(GTK) and PLATFORM(X11). Testing GDK_WINDOWING_X11
ensures that the GTK+ dependency has X11 support, but does not ensure
that WebKitGTK+ has been configured to build for X11 environments (which
is what PLATFORM(X11) ensures).
- 1:07 AM Changeset in webkit [181788] by
-
- 2 edits in releases/WebKitGTK/webkit-2.8/Source/WebCore
Merge r181787 - [GTK] Crash due to empty drag image during drag-and-drop
https://bugs.webkit.org/show_bug.cgi?id=142671
Reviewed by Philippe Normand.
Return early from ImageBuffer constructor if an empty size is
given. This is a speculative fix for a crash while starting a drag
and drop operation, that I haven't been able to reproduce.
- platform/graphics/cairo/ImageBufferCairo.cpp:
(WebCore::ImageBuffer::ImageBuffer):
- 1:04 AM Changeset in webkit [181787] by
-
- 2 edits in trunk/Source/WebCore
[GTK] Crash due to empty drag image during drag-and-drop
https://bugs.webkit.org/show_bug.cgi?id=142671
Reviewed by Philippe Normand.
Return early from ImageBuffer constructor if an empty size is
given. This is a speculative fix for a crash while starting a drag
and drop operation, that I haven't been able to reproduce.
- platform/graphics/cairo/ImageBufferCairo.cpp:
(WebCore::ImageBuffer::ImageBuffer):
- 1:02 AM Changeset in webkit [181786] by
-
- 3 edits in trunk
Unreviewed, kick the GTK bots to regenerate makefile.
https://bugs.webkit.org/show_bug.cgi?id=137394
- Source/cmake/OptionsEfl.cmake: Revert my last accidenatal change.
- Source/cmake/OptionsGTK.cmake: Really kick the GTK bots.
- 12:52 AM Changeset in webkit [181785] by
-
- 2 edits in trunk
Unreviewed, kick the GTK bots to regenerate makefile.
https://bugs.webkit.org/show_bug.cgi?id=137394
- Source/cmake/OptionsEfl.cmake:
