WebKit-https.git
5 years ago[Cocoa] Purge SQLite page cache when under memory pressure.
akling@apple.com [Sun, 1 Mar 2015 02:31:16 +0000 (02:31 +0000)]
[Cocoa] Purge SQLite page cache when under memory pressure.
<https://webkit.org/b/142139>
<rdar://problem/19997739>

Reviewed by Pratik Solanki.

Call out to sqlite3 cache purging SPI on Cocoa platforms when
we need to free up some extra memory.

* platform/cocoa/MemoryPressureHandlerCocoa.mm:
(WebCore::MemoryPressureHandler::platformReleaseMemory):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180847 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoFrameView::layoutTimerFired() should update style if needed before doing layout
simon.fraser@apple.com [Sun, 1 Mar 2015 01:04:05 +0000 (01:04 +0000)]
FrameView::layoutTimerFired() should update style if needed before doing layout
https://bugs.webkit.org/show_bug.cgi?id=141688

Reviewed by Andreas Kling.

If the style recalc timer has been scheduled to fire after the layout timer,
when the layout timer fires, we might as well just do the style recalc
too. The call to updateStyleIfNeeded() will cancel the pending style
recalc timer.

This doesn't have much impact on the number of layouts (measured via PLT)
but seems like a reasonable thing to do.

* page/FrameView.cpp:
(WebCore::FrameView::layoutTimerFired):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180846 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[iOS] Stop running webkit-build-directory on every layout test
ddkilzer@apple.com [Sat, 28 Feb 2015 23:06:20 +0000 (23:06 +0000)]
[iOS] Stop running webkit-build-directory on every layout test

This was originally fixed by David Farler for Bug 135409 in
r172602.

* Scripts/webkitpy/port/ios.py:
(IOSSimulatorPort.relay_path): Mark as @memoized so it doesn't
run webkit-build-directory every time it's called.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180845 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoW3C importer should use filesystem instead of shutil/host
youenn.fablet@crf.canon.fr [Sat, 28 Feb 2015 22:47:40 +0000 (22:47 +0000)]
W3C importer should use filesystem instead of shutil/host
https://bugs.webkit.org/show_bug.cgi?id=142012

Reviewed by Bem Jones-Bey.

Removed direct use of python shutil and os, except for os.walk which will require its own fix.

* Scripts/webkitpy/w3c/test_importer.py:
(main):
(TestImporter.do_import):
(TestImporter.find_importable_tests):
(TestImporter.import_tests):
(TestImporter.remove_deleted_files):
(TestImporter.write_import_log):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180844 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoREGRESSION(r85798): Lists of crashing/timeouting/stderr tests aren't sorted
ossy@webkit.org [Sat, 28 Feb 2015 22:33:34 +0000 (22:33 +0000)]
REGRESSION(r85798): Lists of crashing/timeouting/stderr tests aren't sorted
https://bugs.webkit.org/show_bug.cgi?id=142081

Reviewed by Ryosuke Niwa.

* fast/harness/results.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180843 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWebsiteDataStore should handle fetching and deleting local storage data
andersca@apple.com [Sat, 28 Feb 2015 22:20:17 +0000 (22:20 +0000)]
WebsiteDataStore should handle fetching and deleting local storage data
https://bugs.webkit.org/show_bug.cgi?id=142137

Reviewed by Sam Weinig.

* UIProcess/Storage/StorageManager.cpp:
(WebKit::StorageManager::deleteEntriesForOrigins):
Add a new function that deletes entries from multiple origins.

* UIProcess/Storage/StorageManager.h:
Add new members.

* UIProcess/WebsiteData/WebsiteDataStore.cpp:
(WebKit::WebsiteDataStore::fetchData):
Fetch local storage data as well.

(WebKit::WebsiteDataStore::removeData):
Delete local storage data as well.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180842 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoFetch cache origins from the network process
andersca@apple.com [Sat, 28 Feb 2015 21:52:59 +0000 (21:52 +0000)]
Fetch cache origins from the network process
https://bugs.webkit.org/show_bug.cgi?id=142135

Reviewed by Dan Bernstein.

* NetworkProcess/NetworkProcess.cpp:
(WebKit::cfURLCacheOrigins):
Add a new helper function that returns a vector of CFURL cache origins.

(WebKit::fetchDiskCacheOrigins):
Fetch the disk cache origins and pass them along to the completion handler. Currently we don't handle the new disk cache.

(WebKit::NetworkProcess::fetchWebsiteData):
Create a callback aggregator and fetch disk cache origins if we're asked for it.

* UIProcess/WebsiteData/WebsiteDataStore.cpp:
(WebKit::computeNetworkProcessAccessTypeForDataFetch):
New helper function that computes if we need to launch the network process in order to fetch data.

(WebKit::WebsiteDataStore::fetchData):
Fetch website data from the network process as well.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180841 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed - set svn:ignore for generated web-platform-tests files - see bug 142110
youenn.fablet@crf.canon.fr [Sat, 28 Feb 2015 21:41:42 +0000 (21:41 +0000)]
Unreviewed - set svn:ignore for generated web-platform-tests files - see bug 142110

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180840 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoFullscreen video layers are off by one sometimes
simon.fraser@apple.com [Sat, 28 Feb 2015 20:58:51 +0000 (20:58 +0000)]
Fullscreen video layers are off by one sometimes
https://bugs.webkit.org/show_bug.cgi?id=142122
rdar://problem/19878821

Reviewed by Eric Carlson.

Convert MediaPlayer::naturalSize() to return a FloatSize, since the natural size
isn't always integral (because of preserving pixel aspect ratio etc). Fix all the media
backends to use FloatSizes for natural size. Convert the video image drawing code
paths to FloatSize, since naturalSize is used on the destination rect computation,
and painting should be floating point anyway.

Give the layer created by SourceBufferPrivateAVFObjC a name in debug builds.

* html/HTMLVideoElement.cpp:
(WebCore::HTMLVideoElement::videoWidth):
(WebCore::HTMLVideoElement::videoHeight):
(WebCore::HTMLVideoElement::paintCurrentFrameInContext):
* html/HTMLVideoElement.h:
* html/canvas/CanvasRenderingContext2D.cpp:
(WebCore::size):
(WebCore::CanvasRenderingContext2D::drawImage):
* html/canvas/WebGLRenderingContextBase.cpp:
(WebCore::WebGLRenderingContextBase::videoFrameToImage):
* platform/graphics/MediaPlayer.cpp:
(WebCore::NullMediaPlayerPrivate::naturalSize):
(WebCore::MediaPlayer::naturalSize):
(WebCore::MediaPlayer::paint):
(WebCore::MediaPlayer::paintCurrentFrameInContext):
(WebCore::NullMediaPlayerPrivate::paint): Deleted.
* platform/graphics/MediaPlayer.h:
* platform/graphics/MediaPlayerPrivate.h:
(WebCore::MediaPlayerPrivateInterface::paintCurrentFrameInContext):
* platform/graphics/avfoundation/MediaPlayerPrivateAVFoundation.cpp:
(WebCore::MediaPlayerPrivateAVFoundation::naturalSize):
(WebCore::MediaPlayerPrivateAVFoundation::setNaturalSize):
* platform/graphics/avfoundation/MediaPlayerPrivateAVFoundation.h:
* platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.h:
* platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.mm:
(WebCore::MediaPlayerPrivateAVFoundationObjC::paintCurrentFrameInContext):
(WebCore::MediaPlayerPrivateAVFoundationObjC::paint):
(WebCore::MediaPlayerPrivateAVFoundationObjC::paintWithImageGenerator):
(WebCore::MediaPlayerPrivateAVFoundationObjC::createImageForTimeInRect):
(WebCore::MediaPlayerPrivateAVFoundationObjC::tracksChanged):
(WebCore::MediaPlayerPrivateAVFoundationObjC::sizeChanged):
(WebCore::MediaPlayerPrivateAVFoundationObjC::paintWithVideoOutput):
* platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.h:
* platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.mm:
(WebCore::MediaPlayerPrivateMediaSourceAVFObjC::naturalSize):
(WebCore::MediaPlayerPrivateMediaSourceAVFObjC::paint):
(WebCore::MediaPlayerPrivateMediaSourceAVFObjC::paintCurrentFrameInContext):
* platform/graphics/avfoundation/objc/MediaSourcePrivateAVFObjC.h:
* platform/graphics/avfoundation/objc/MediaSourcePrivateAVFObjC.mm:
(WebCore::MediaSourcePrivateAVFObjC::naturalSize):
* platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.h:
* platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:
(WebCore::SourceBufferPrivateAVFObjC::trackDidChangeEnabled):
(WebCore::SourceBufferPrivateAVFObjC::naturalSize):
* platform/graphics/avfoundation/objc/VideoTrackPrivateMediaSourceAVFObjC.h:
* platform/graphics/avfoundation/objc/VideoTrackPrivateMediaSourceAVFObjC.mm:
(WebCore::VideoTrackPrivateMediaSourceAVFObjC::naturalSize):
* platform/graphics/mac/MediaPlayerPrivateQTKit.h:
* platform/graphics/mac/MediaPlayerPrivateQTKit.mm:
(WebCore::MediaPlayerPrivateQTKit::naturalSize):
(WebCore::MediaPlayerPrivateQTKit::paintCurrentFrameInContext):
(WebCore::MediaPlayerPrivateQTKit::paint):
* platform/mock/mediasource/MockMediaPlayerMediaSource.cpp:
(WebCore::MockMediaPlayerMediaSource::naturalSize):
(WebCore::MockMediaPlayerMediaSource::paint):
* platform/mock/mediasource/MockMediaPlayerMediaSource.h:
* rendering/RenderVideo.cpp:
(WebCore::RenderVideo::calculateIntrinsicSize):
(WebCore::RenderVideo::paintReplaced):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180839 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoSimplify WebResourceCacheManagerCFNet.mm code
andersca@apple.com [Sat, 28 Feb 2015 20:20:31 +0000 (20:20 +0000)]
Simplify WebResourceCacheManagerCFNet.mm code
https://bugs.webkit.org/show_bug.cgi?id=142134

Reviewed by Dan Bernstein.

- ENABLE(CACHE_PARTITIONING) is always true on Mac and iOS, so remove those #ifdefs.
- Make cfURLCacheHostNamesWithCallback and clearCFURLCacheForHostNames public
  so they can be used by the new WebsiteDataStore code in an upcoming patch.
- Change cfURLCacheHostNamesWithCallback to take an std::function instead of a block.

* WebProcess/ResourceCache/WebResourceCacheManager.cpp:
(WebKit::WebResourceCacheManager::getCacheOrigins):
* WebProcess/ResourceCache/WebResourceCacheManager.h:
* WebProcess/ResourceCache/cf/WebResourceCacheManagerCFNet.mm:
(WebKit::partitionName):
(WebKit::WebResourceCacheManager::cfURLCacheHostNamesWithCallback):
(WebKit::WebResourceCacheManager::clearCFURLCacheForHostNames):
(WebKit::WebResourceCacheManager::cfURLCacheHostNames): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180838 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed, rolling out r180804.
commit-queue@webkit.org [Sat, 28 Feb 2015 18:45:42 +0000 (18:45 +0000)]
Unreviewed, rolling out r180804.
https://bugs.webkit.org/show_bug.cgi?id=142131

Broke nightlies (Requested by ap on #webkit).

Reverted changeset:

"[WK2] Drop legacy WKBundlePageDiagnosticLoggingClient API"
https://bugs.webkit.org/show_bug.cgi?id=141176
http://trac.webkit.org/changeset/180804

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180837 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed EFL gardening 1st Mar.
gyuyoung.kim@samsung.com [Sat, 28 Feb 2015 15:05:51 +0000 (15:05 +0000)]
Unreviewed EFL gardening 1st Mar.

* platform/efl/TestExpectations:
- Unskip tests which have been passed since r180672.
- Skip compositing/webgl.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180836 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoW3C test importer should use filesystem to read and write files
youenn.fablet@crf.canon.fr [Sat, 28 Feb 2015 11:49:20 +0000 (11:49 +0000)]
W3C test importer should use filesystem to read and write files
https://bugs.webkit.org/show_bug.cgi?id=142084

Reviewed by Bem Jones-Bey.

Use of FileSystem.write_binary_file, read_text_file and write_text_file in lieu of open().

* Scripts/webkitpy/w3c/test_importer.py:
(TestImporter.import_tests):
(TestImporter.remove_deleted_files):
(TestImporter.write_import_log):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180830 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoSource/WebCore:
ossy@webkit.org [Sat, 28 Feb 2015 07:26:02 +0000 (07:26 +0000)]
Source/WebCore:
[EFL][GTK] Fix build break after r180790,180798
https://bugs.webkit.org/show_bug.cgi?id=142127

Patch by Hunseop Jeong <hs85.jeong@samsung.com> on 2015-02-27
Reviewed by Gyuyoung Kim.

* platform/graphics/cairo/PathCairo.cpp:
(WebCore::Path::addEllipse):

Source/WebKit2:
[EFL][GTK] Fix build break after r180790,180798
https://bugs.webkit.org/show_bug.cgi?id=142127

Patch by Hunseop Jeong <hs85.jeong@samsung.com> on 2015-02-27
Reviewed by Gyuyoung Kim.

* CMakeLists.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180817 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Win] Unreviewed build fix.
bfulgham@apple.com [Sat, 28 Feb 2015 07:12:47 +0000 (07:12 +0000)]
[Win] Unreviewed build fix.

Adjust project dependencies to ensure a consistent build order.

* WebKit.vcxproj/WebKit.sln:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180816 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoSubpixel-layout: width: max-content; property might cause unnecessary scrollbar.
zalan@apple.com [Sat, 28 Feb 2015 06:21:05 +0000 (06:21 +0000)]
Subpixel-layout: width: max-content; property might cause unnecessary scrollbar.
https://bugs.webkit.org/show_bug.cgi?id=142065

Reviewed by Simon Fraser.

Source/WebCore:

We should not pixelsnap (ceil in this case) logical coordinates during layout.
Should this cause content to be partially cut off, we need to
find the broken piece in the computation logic.

Covered by the unskipped test.

* rendering/RootInlineBox.cpp:
(WebCore::RootInlineBox::paddedLayoutOverflowRect):

LayoutTests:

Rebaseline. Scroll layer shrinks in certain cases.

* platform/mac-mavericks/editing/input/caret-at-the-edge-of-input-expected.txt:
* platform/mac-mavericks/fast/css/text-overflow-input-expected.txt:
* platform/mac-mavericks/fast/forms/basic-inputs-expected.txt:
* platform/mac-mavericks/fast/forms/control-restrict-line-height-expected.txt:
* platform/mac-mavericks/fast/forms/input-disabled-color-expected.txt:
* platform/mac-mavericks/fast/forms/search/search-size-with-decorations-expected.txt:
* platform/mac-mavericks/http/tests/navigation/javascriptlink-frames-expected.txt:
* platform/mac/TestExpectations:
* platform/mac/fast/forms/control-restrict-line-height-expected.txt:
* platform/mac/fast/forms/input-appearance-selection-expected.txt:
* platform/mac/fast/forms/input-text-scroll-left-on-blur-expected.txt:
* platform/mac/fast/forms/input-type-text-min-width-expected.txt:
* platform/mac/fast/forms/minWidthPercent-expected.txt:
* platform/mac/fast/forms/search/search-size-with-decorations-expected.txt:
* platform/mac/tables/mozilla/bugs/bug96334-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180815 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoRemove unused PossiblyNull
darin@apple.com [Sat, 28 Feb 2015 06:02:30 +0000 (06:02 +0000)]
Remove unused PossiblyNull
https://bugs.webkit.org/show_bug.cgi?id=142124

Reviewed by Andreas Kling.

* WTF.vcxproj/WTF.vcxproj: Removed the file.
* WTF.vcxproj/WTF.vcxproj.filters: Ditto.
* WTF.xcodeproj/project.pbxproj: Ditto.
* wtf/CMakeLists.txt: Ditto.
* wtf/PossiblyNull.h: Removed.

* wtf/FastMalloc.h: Moved everything to the left.
Moved member functions out of the TryMallocReturnValue class definition.
(WTF::TryMallocReturnValue::operator PossiblyNull<T>): Deleted.
(WTF::TryMallocReturnValue::getValue): Marked inline, changed to work
only with pointer types, not arbitrary non-pointer types.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180814 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[JSC] Use the way number constants are written to help type speculation
benjamin@webkit.org [Sat, 28 Feb 2015 03:21:37 +0000 (03:21 +0000)]
[JSC] Use the way number constants are written to help type speculation
https://bugs.webkit.org/show_bug.cgi?id=142072

Patch by Benjamin Poulain <bpoulain@apple.com> on 2015-02-27
Reviewed by Filip Pizlo.

This patch changes how we interpret numeric constant based on how they appear
in the source.

Constants that are integers but written with a decimal point now carry that information
to the optimizating tiers. From there, we use that to be more aggressive about typing
math operations toward double operations.

For example, in:
    var a = x + 1.0;
    var b = y + 1;
The Add for a would be biased toward doubles, the Add for b would speculate
integer as usual.

The gains are tiny but this is a prerequisite to make my next patch useful:
-SunSpider's access-fannkuch: definitely 1.0661x faster
-SunSpider's math-cordic: definitely 1.0266x slower
    overal: might be 1.0066x slower.
-Kraken's imaging-darkroom: definitely 1.0333x faster.

* parser/Lexer.cpp:
(JSC::tokenTypeForIntegerLikeToken):
(JSC::Lexer<T>::lex):
The lexer now create two types of tokens for number: INTEGER and DOUBLE.
Those token types only carry information about how the values were
entered, an INTEGER does not have to be an integer, it is only written like one.
Large integer still end up represented as double in memory.

One trap I fell into was typing numbers like 12e3 as double. This kind of literal
is frequently used in integer-typed code, while 12.e3 would appear in double-typed
code.
Because of that, the only signals for double are: decimal point, negative zero,
and ridiculously large values.

* parser/NodeConstructors.h:
(JSC::DoubleNode::DoubleNode):
(JSC::IntegerNode::IntegerNode):
* parser/Nodes.h:
(JSC::NumberNode::value):
(JSC::NumberNode::setValue): Deleted.
Number get specialized in two new kind of nodes in the AST: IntegerNode and DoubleNode.

* bytecompiler/NodesCodegen.cpp:
(JSC::NumberNode::emitBytecode):

* parser/ASTBuilder.h:
(JSC::ASTBuilder::createDoubleExpr):
(JSC::ASTBuilder::createIntegerExpr):
(JSC::ASTBuilder::createIntegerLikeNumber):
(JSC::ASTBuilder::createDoubleLikeNumber):
(JSC::ASTBuilder::createNumberFromBinaryOperation):
(JSC::ASTBuilder::createNumberFromUnaryOperation):
(JSC::ASTBuilder::makeNegateNode):
(JSC::ASTBuilder::makeBitwiseNotNode):
(JSC::ASTBuilder::makeMultNode):
(JSC::ASTBuilder::makeDivNode):
(JSC::ASTBuilder::makeModNode):
(JSC::ASTBuilder::makeAddNode):
(JSC::ASTBuilder::makeSubNode):
(JSC::ASTBuilder::makeLeftShiftNode):
(JSC::ASTBuilder::makeRightShiftNode):
(JSC::ASTBuilder::makeURightShiftNode):
(JSC::ASTBuilder::makeBitOrNode):
(JSC::ASTBuilder::makeBitAndNode):
(JSC::ASTBuilder::makeBitXOrNode):
(JSC::ASTBuilder::createNumberExpr): Deleted.
(JSC::ASTBuilder::createNumber): Deleted.
The AST has some optimization to resolve constants before emitting bytecode.
In the new code, the intger representation is kept if both operands where
also represented as integers.

* parser/Parser.cpp:
(JSC::Parser<LexerType>::parseDeconstructionPattern):
(JSC::Parser<LexerType>::parseProperty):
(JSC::Parser<LexerType>::parseGetterSetter):
(JSC::Parser<LexerType>::parsePrimaryExpression):
(JSC::Parser<LexerType>::printUnexpectedTokenText):
* parser/ParserTokens.h:
* parser/SyntaxChecker.h:
(JSC::SyntaxChecker::createDoubleExpr):
(JSC::SyntaxChecker::createIntegerExpr):
(JSC::SyntaxChecker::createNumberExpr): Deleted.

* bytecode/CodeBlock.cpp:
(JSC::CodeBlock::registerName):
(JSC::CodeBlock::constantName):
Change constantName(r, getConstant(r)) -> constantName(r) to simplify
the dump code.

(JSC::CodeBlock::dumpBytecode):
Dump thre soure representation information we have with each constant.

(JSC::CodeBlock::CodeBlock):
(JSC::CodeBlock::shrinkToFit):
(JSC::constantName): Deleted.
* bytecode/CodeBlock.h:
(JSC::CodeBlock::constantsSourceCodeRepresentation):
(JSC::CodeBlock::addConstant):
(JSC::CodeBlock::addConstantLazily):
(JSC::CodeBlock::constantSourceCodeRepresentation):
(JSC::CodeBlock::setConstantRegisters):

* bytecode/UnlinkedCodeBlock.h:
(JSC::UnlinkedCodeBlock::addConstant):
(JSC::UnlinkedCodeBlock::constantsSourceCodeRepresentation):
(JSC::UnlinkedCodeBlock::shrinkToFit):

* bytecompiler/BytecodeGenerator.cpp:
(JSC::BytecodeGenerator::addConstantValue):
(JSC::BytecodeGenerator::emitLoad):
* bytecompiler/BytecodeGenerator.h:
We have to differentiate between constants that have the same values but are
represented differently in the source. Values like 1.0 and 1 now end up
as different constants.

* dfg/DFGByteCodeParser.cpp:
(JSC::DFG::ByteCodeParser::get):
(JSC::DFG::ByteCodeParser::addConstantToGraph):
* dfg/DFGGraph.cpp:
(JSC::DFG::Graph::registerFrozenValues):
* dfg/DFGGraph.h:
(JSC::DFG::Graph::addSpeculationMode):
(JSC::DFG::Graph::addImmediateShouldSpeculateInt32):
ArithAdd is very aggressive toward using Int52, which is quite useful
in many benchmarks.

Here we need to specialize to make sure we don't force our literals
to Int52 if there were represented as double.

There is one exception to that rule: when the other operand is guaranteed
to come from a NodeResultInt32. This is because there is some weird code
doing stuff like:
    var b = a|0;
    var c = b*2.0;

* dfg/DFGNode.h:
(JSC::DFG::Node::Node):
(JSC::DFG::Node::setOpAndDefaultFlags):
(JSC::DFG::Node::sourceCodeRepresentation):
* dfg/DFGPredictionPropagationPhase.cpp:
(JSC::DFG::PredictionPropagationPhase::propagate):
* runtime/JSCJSValue.h:
(JSC::EncodedJSValueWithRepresentationHashTraits::emptyValue):
(JSC::EncodedJSValueWithRepresentationHashTraits::constructDeletedValue):
(JSC::EncodedJSValueWithRepresentationHashTraits::isDeletedValue):
(JSC::EncodedJSValueWithRepresentationHash::hash):
(JSC::EncodedJSValueWithRepresentationHash::equal):
* tests/stress/arith-add-with-constants.js: Added.
* tests/stress/arith-mul-with-constants.js: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180813 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[iOS] Gardening: rebaseline fast/attachment results
ddkilzer@apple.com [Sat, 28 Feb 2015 03:16:25 +0000 (03:16 +0000)]
[iOS] Gardening: rebaseline fast/attachment results

* platform/ios-simulator/fast/attachment/attachment-disabled-dom-expected.txt: Add.
* platform/ios-simulator/fast/attachment/attachment-disabled-rendering-expected.txt: Update.
* platform/ios-simulator/fast/attachment/attachment-dom-expected.txt: Add.
* platform/ios-simulator/fast/attachment/attachment-rendering-expected.txt: Update.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180812 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[iOS] Gardening: Unskip fast/history/page-cache-webdatabase-opened-db.html
ddkilzer@apple.com [Sat, 28 Feb 2015 02:51:18 +0000 (02:51 +0000)]
[iOS] Gardening: Unskip fast/history/page-cache-webdatabase-opened-db.html

* platform/ios-simulator/TestExpectations: Unskip test.  It
passes on iOS WK1 and WK2 because WebSQL != IndexedDB.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180811 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[iOS] Gardening: Skip some new page-cache tests
ddkilzer@apple.com [Sat, 28 Feb 2015 02:51:14 +0000 (02:51 +0000)]
[iOS] Gardening: Skip some new page-cache tests

* platform/ios-simulator/TestExpectations:
- Skip MediaSource tests:
  fast/history/page-cache-media-source-closed-2.html
  fast/history/page-cache-media-source-closed.html
  fast/history/page-cache-media-source-opened.html
- Skip test that uses drag-and-drop:
  fast/history/page-cache-createObjectURL.html

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180810 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoNode::hasEditableStyle and isEditablePosition have too many options
rniwa@webkit.org [Sat, 28 Feb 2015 02:41:12 +0000 (02:41 +0000)]
Node::hasEditableStyle and isEditablePosition have too many options
https://bugs.webkit.org/show_bug.cgi?id=142078

Reviewed by Andreas Kling.

Moved the code that dealt with accessibility to htmlediting.cpp from Node. This patch introduces
new editing helper functions hasEditableStyle and isEditableNode for this purpose.

Also removed UserSelectAllTreatment from isContentEditable's arguments in the favor of using
newly extracted computeEditability in call sites that specify this option since isContentEditable
is a public DOM API.

No new tests since there should be no observable behavior changes.

* accessibility/AXObjectCache.h: Removed the declaration of an undefined function.

* dom/Element.cpp:
(WebCore::Element::shouldUseInputMethod): Uses newly added computeEditability.

* dom/Node.cpp:
(WebCore::Node::isContentEditable): Ditto. No longer takes UserSelectAllTreatment as an argument.
(WebCore::Node::isContentRichlyEditable): Ditto.
(WebCore::Node::computeEditability): Renamed from hasEditableStyle to avoid the confusion with
a helper function of the same name. Added ShouldUpdateStyle as an argument to optionally update
style tree. Also returns tri-state Editability enum instead of returning a boolean based on
the value of EditableLevel argument.
(WebCore::Node::isEditableToAccessibility): Moved to htmlediting.cpp.
(WebCore::Node::willRespondToMouseClickEvents): Uses newly added computeEditability.
(WebCore::Node::rootEditableElement): Moved to htmlediting.cpp.

* dom/Node.h: No longer includes EditingBoundary.h.
(WebCore::Node::isContentEditable):
(WebCore::Node::hasEditableStyle): No longer takes EditableType as an argument.
(WebCore::Node::hasRichlyEditableStyle): Ditto.

* editing/ApplyStyleCommand.cpp:
(WebCore::ApplyStyleCommand::removeInlineStyleFromElement): Uses newly added isEditableNode.
(WebCore::ApplyStyleCommand::surroundNodeRangeWithElement): Ditto.
* editing/DeleteFromTextNodeCommand.cpp:
(WebCore::DeleteFromTextNodeCommand::doApply): Ditto.
* editing/FrameSelection.cpp:
(WebCore::CaretBase::invalidateCaretRect): Ditto.
* editing/InsertNodeBeforeCommand.cpp:
(WebCore::InsertNodeBeforeCommand::doApply): Ditto.
(WebCore::InsertNodeBeforeCommand::doUnapply): Ditto.
* editing/RemoveNodeCommand.cpp:
(WebCore::RemoveNodeCommand::doApply): Ditto.

* editing/VisibleSelection.cpp:
(WebCore::VisibleSelection::hasEditableStyle): Since this is the only caller of isEditablePosition
which sets DoNotUpdateStyle, directly call hasEditableStyle on the container node instead. This was
not possible prior to r180726 because isEditablePosition had to move out of tables.

* editing/VisibleUnits.cpp:
(WebCore::previousLeafWithSameEditability): Uses newly added hasEditableStyle.
(WebCore::nextLeafWithSameEditability): Ditto.
(WebCore::rootEditableOrDocumentElement): Extracted from previousLinePosition. Use helper functions
in htmlediting.cpp instead of member functions of Node since they no longer support EditableType.
(WebCore::previousLinePosition):
(WebCore::nextLinePosition):

* editing/htmlediting.cpp:
(WebCore::highestEditableRoot): Uses newly added hasEditableStyle.
(WebCore::isEditableToAccessibility): Moved from Node.
(WebCore::computeEditability): Extracted from isEditablePosition.
(WebCore::hasEditableStyle): Added.
(WebCore::isEditableNode): Added.
(WebCore::isEditablePosition): Now calls computeEditability.
(WebCore::isRichlyEditablePosition): No longer takes EditableType since that variant was never used.
(WebCore::editableRootForPosition): Moved the code from Node::rootEditableElement.

* editing/htmlediting.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180809 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoTest gardening for Windows after r180796.
mmaxfield@apple.com [Sat, 28 Feb 2015 02:14:24 +0000 (02:14 +0000)]
Test gardening for Windows after r180796.

Unreviewed.

Patch by Myles C. Maxfield <mmaxfield@apple.com> on 2015-02-27

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180808 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agohttps://bugs.webkit.org/show_bug.cgi?id=142119
dino@apple.com [Sat, 28 Feb 2015 01:57:46 +0000 (01:57 +0000)]
https://bugs.webkit.org/show_bug.cgi?id=142119

Unreviewed. Add fast/canvas/canvas-ellipse-zero-lineto.html
to list of failing tests on Windows.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180807 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoFixed a typo in the previous commit.
rniwa@webkit.org [Sat, 28 Feb 2015 01:52:48 +0000 (01:52 +0000)]
Fixed a typo in the previous commit.

* bmalloc/BoundaryTag.h:
(bmalloc::BoundaryTag::setOwner):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180806 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoEFL build fix after r180797.
rniwa@webkit.org [Sat, 28 Feb 2015 01:46:05 +0000 (01:46 +0000)]
EFL build fix after r180797.

* bmalloc/BoundaryTag.h:
(bmalloc::BoundaryTag::owner):
(bmalloc::BoundaryTag::setOwner):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180805 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[WK2] Drop legacy WKBundlePageDiagnosticLoggingClient API
cdumez@apple.com [Sat, 28 Feb 2015 01:43:24 +0000 (01:43 +0000)]
[WK2] Drop legacy WKBundlePageDiagnosticLoggingClient API
https://bugs.webkit.org/show_bug.cgi?id=141176

Reviewed by Sam Weinig.

Drop legacy WKBundlePageDiagnosticLoggingClient WK2 API as we are now
exposing this functionality via WKPageDiagnosticLoggingClient.h on
UIProcess side. The client-side has already been ported over.

* CMakeLists.txt:
* Shared/API/c/WKSharedAPICast.h:
(WebKit::toAPI): Deleted.
(WebKit::toDiagnosticLoggingResultType): Deleted.
* UIProcess/API/C/WKAPICast.h:
(WebKit::toAPI):
* UIProcess/API/C/WKDiagnosticLoggingResultType.h: Renamed from Source/WebKit2/Shared/API/c/WKDiagnosticLoggingResultType.h.
* WebKit2.xcodeproj/project.pbxproj:
* WebProcess/InjectedBundle/API/c/WKBundlePage.cpp:
(WKBundlePageSetDiagnosticLoggingClient): Deleted.
* WebProcess/InjectedBundle/API/c/WKBundlePage.h:
* WebProcess/InjectedBundle/API/c/WKBundlePageDiagnosticLoggingClient.h: Removed.
* WebProcess/InjectedBundle/InjectedBundlePageDiagnosticLoggingClient.cpp: Removed.
* WebProcess/InjectedBundle/InjectedBundlePageDiagnosticLoggingClient.h: Removed.
* WebProcess/WebCoreSupport/WebDiagnosticLoggingClient.cpp:
(WebKit::WebDiagnosticLoggingClient::logDiagnosticMessage):
(WebKit::WebDiagnosticLoggingClient::logDiagnosticMessageWithResult):
(WebKit::WebDiagnosticLoggingClient::logDiagnosticMessageWithValue):
* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::WebPage):
(WebKit::WebPage::close):
(WebKit::WebPage::initializeInjectedBundleDiagnosticLoggingClient): Deleted.
* WebProcess/WebPage/WebPage.h:
(WebKit::WebPage::injectedBundleDiagnosticLoggingClient): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180804 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoRename WebResourceCacheManagerCFNet.cpp to WebResourceCacheManagerCFNet.mm
andersca@apple.com [Sat, 28 Feb 2015 01:19:06 +0000 (01:19 +0000)]
Rename WebResourceCacheManagerCFNet.cpp to WebResourceCacheManagerCFNet.mm

Rubber-stamped by Dan Bernstein.

This will let us use lambda to block conversion in a subsequent patch.

* WebKit2.xcodeproj/project.pbxproj:
* WebProcess/ResourceCache/cf/WebResourceCacheManagerCFNet.mm: Renamed from Source/WebKit2/WebProcess/ResourceCache/cf/WebResourceCacheManagerCFNet.cpp.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180802 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoMake ActiveDOMObject::canSuspend() pure virtual
cdumez@apple.com [Sat, 28 Feb 2015 01:05:21 +0000 (01:05 +0000)]
Make ActiveDOMObject::canSuspend() pure virtual
https://bugs.webkit.org/show_bug.cgi?id=142096
<rdar://problem/19923085>

Reviewed by Andreas Kling.

Make ActiveDOMObject::canSuspend() pure virtual so that people at least
try to provide an implementation for it. The default implementation was
returning false unconditionally and thus was preventing pages from
entering the PageCache.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180801 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed, rolling out r180203 and r180210.
commit-queue@webkit.org [Sat, 28 Feb 2015 00:46:33 +0000 (00:46 +0000)]
Unreviewed, rolling out r180203 and r180210.
https://bugs.webkit.org/show_bug.cgi?id=142116

broke process suspension and tile map (Requested by thorton on
#webkit).

Reverted changesets:

"Adopt CAMachPort-as-layer-contents"
https://bugs.webkit.org/show_bug.cgi?id=141687
http://trac.webkit.org/changeset/180203

"Fix the !USE(IOSURFACE) build"
http://trac.webkit.org/changeset/180210

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180799 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdd WebKit2 SPI to create a DOM File object
commit-queue@webkit.org [Sat, 28 Feb 2015 00:42:26 +0000 (00:42 +0000)]
Add WebKit2 SPI to create a DOM File object
https://bugs.webkit.org/show_bug.cgi?id=142109

Patch by Sam Weinig <sam@webkit.org> on 2015-02-27
Reviewed by Tim Horton.

Source/WebCore:

* WebCore.xcodeproj/project.pbxproj:
Make <WebCore/File.h> (and associated files) available to WebKit2.

Source/WebKit2:

Add a new handle type for exposing a DOM File object to script. Follow
the pattern of WKBundleNodeHandleRef and WKBundleRangeHandleRef with the
new class WKBundleFileHandleRef. It can be created for a specific path,
and then the JS wrapper can be obtained via WKBundleFrameGetJavaScriptWrapperForFileForWorld.

* Shared/API/APIObject.h:
* Shared/API/c/WKBase.h:
* WebKit2.xcodeproj/project.pbxproj:
* WebProcess/InjectedBundle/API/c/WKBundleAPICast.h:
* WebProcess/InjectedBundle/API/c/WKBundleFileHandleRef.cpp: Added.
(WKBundleFileHandleGetTypeID):
(WKBundleFileHandleCreateWithPath):
* WebProcess/InjectedBundle/API/c/WKBundleFileHandleRef.h: Added.
* WebProcess/InjectedBundle/API/c/WKBundleFrame.cpp:
(WKBundleFrameGetJavaScriptWrapperForFileForWorld):
* WebProcess/InjectedBundle/API/c/WKBundleFrame.h:
* WebProcess/InjectedBundle/DOM/InjectedBundleFileHandle.cpp: Added.
(WebKit::domHandleCache):
(WebKit::InjectedBundleFileHandle::create):
(WebKit::InjectedBundleFileHandle::getOrCreate):
(WebKit::InjectedBundleFileHandle::InjectedBundleFileHandle):
(WebKit::InjectedBundleFileHandle::~InjectedBundleFileHandle):
(WebKit::InjectedBundleFileHandle::coreFile):
* WebProcess/InjectedBundle/DOM/InjectedBundleFileHandle.h: Added.
* WebProcess/WebPage/WebFrame.cpp:
(WebKit::WebFrame::jsWrapperForWorld):
* WebProcess/WebPage/WebFrame.h:

Tools:

Add a test for WKBundleFileHandleRef.

* TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
* TestWebKitAPI/Tests/WebKit2/WKBundleFileHandle.cpp: Added.
(TestWebKitAPI::didReceiveMessageFromInjectedBundle):
(TestWebKitAPI::didFinishLoadForFrame):
(TestWebKitAPI::TEST):
* TestWebKitAPI/Tests/WebKit2/WKBundleFileHandle_Bundle.cpp: Added.
(TestWebKitAPI::WKBundleFileHandleTest::WKBundleFileHandleTest):
* TestWebKitAPI/Tests/WebKit2/bundle-file.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180798 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agobmalloc: Pathological madvise churn on the free(malloc(x)) benchmark
ggaren@apple.com [Sat, 28 Feb 2015 00:29:22 +0000 (00:29 +0000)]
bmalloc: Pathological madvise churn on the free(malloc(x)) benchmark
https://bugs.webkit.org/show_bug.cgi?id=142058

Reviewed by Andreas Kling.

The churn was caused by repeatedly splitting an object with physical
pages from an object without, and then merging them back together again.
The merge would conservatively forget that we had physical pages, forcing
a new call to madvise on the next allocation.

This patch more strictly segregates objects in the heap from objects in
the VM heap, with these changes:

(1) Objects in the heap are not allowed to merge with objects in the VM
heap, and vice versa -- since that would erase our precise knowledge of
which physical pages had been allocated.

(2) The VM heap is exclusively responsible for allocating and deallocating
physical pages.

(3) The heap free list must consider entries for objects that are in the
VM heap to be invalid, and vice versa. (This condition can arise
because the free list does not eagerly remove items.)

With these changes, we can know that any valid object in the heap's free
list already has physical pages, and does not need to call madvise.

Note that the VM heap -- as before -- might sometimes contain ranges
or pieces of ranges that have physical pages, since we allow splitting
of ranges at granularities smaller than the VM page size. These ranges
can eventually merge with ranges in the heap during scavenging.

* bmalloc.xcodeproj/project.pbxproj:

* bmalloc/BoundaryTag.h:
(bmalloc::BoundaryTag::owner):
(bmalloc::BoundaryTag::setOwner):
(bmalloc::BoundaryTag::initSentinel):
(bmalloc::BoundaryTag::hasPhysicalPages): Deleted.
(bmalloc::BoundaryTag::setHasPhysicalPages): Deleted. Replaced the concept
of "has physical pages" with a bit indicating which heap owns the large
object. This is a more precise concept, since the old bit was really a
Yes / Maybe bit.

* bmalloc/Deallocator.cpp:

* bmalloc/FreeList.cpp: Adopt
(bmalloc::FreeList::takeGreedy):
(bmalloc::FreeList::take):
(bmalloc::FreeList::removeInvalidAndDuplicateEntries):
* bmalloc/FreeList.h:
(bmalloc::FreeList::push): Added API for considering the owner when
deciding if a free list entry is valid.

* bmalloc/Heap.cpp:
(bmalloc::Heap::Heap): Adopt new API.

(bmalloc::Heap::scavengeLargeRanges): Scavenge all ranges with no minimum,
since some ranges might be able to merge with ranges in the VM heap, and
they won't be allowed to until we scavenge them.

(bmalloc::Heap::allocateSmallPage):
(bmalloc::Heap::allocateMediumPage):
(bmalloc::Heap::allocateLarge): New VM heap API makes this function
simpler, since we always get back physical pages now.

* bmalloc/Heap.h:
* bmalloc/LargeObject.h:
(bmalloc::LargeObject::end):
(bmalloc::LargeObject::owner):
(bmalloc::LargeObject::setOwner):
(bmalloc::LargeObject::isValidAndFree):
(bmalloc::LargeObject::merge): Do not merge objects across heaps since
that causes madvise churn.
(bmalloc::LargeObject::validateSelf):
(bmalloc::LargeObject::init):
(bmalloc::LargeObject::hasPhysicalPages): Deleted.
(bmalloc::LargeObject::setHasPhysicalPages): Deleted. Propogate the Owner API.

* bmalloc/Owner.h: Added.

* bmalloc/SegregatedFreeList.cpp:
(bmalloc::SegregatedFreeList::SegregatedFreeList):
(bmalloc::SegregatedFreeList::insert):
(bmalloc::SegregatedFreeList::takeGreedy):
(bmalloc::SegregatedFreeList::take):
* bmalloc/SegregatedFreeList.h: Propogate the owner API.

* bmalloc/VMAllocate.h:
(bmalloc::vmDeallocatePhysicalPagesSloppy):
(bmalloc::vmAllocatePhysicalPagesSloppy): Clarified these functions and
removed an edge case.

* bmalloc/VMHeap.cpp:
(bmalloc::VMHeap::VMHeap):
* bmalloc/VMHeap.h:
(bmalloc::VMHeap::allocateSmallPage):
(bmalloc::VMHeap::allocateMediumPage):
(bmalloc::VMHeap::allocateLargeObject):
(bmalloc::VMHeap::deallocateLargeObject): Be sure to give each object
a new chance to merge, since it might have been prohibited from merging
before by virtue of not being in the VM heap.

(bmalloc::VMHeap::allocateLargeRange): Deleted.
(bmalloc::VMHeap::deallocateLargeRange): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180797 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUpdating more tests after r177774
mmaxfield@apple.com [Sat, 28 Feb 2015 00:28:12 +0000 (00:28 +0000)]
Updating more tests after r177774

Unreviewed.

Most of these tests simply need to be updated. However, I found two real bugs while
going through these!

* fast/css-generated-content/after-with-inline-continuation-expected.html:
* fast/css-generated-content/after-with-inline-continuation.html:
* fast/inline/hidpi-pixel-gap-between-adjacent-selection-inlines-expected.html:
* fast/inline/hidpi-select-inline-on-subpixel-position-expected.html:
* fast/inline/hidpi-select-inline-on-subpixel-position.html:
* fast/lists/rtl-marker-expected.html:
* fast/lists/rtl-marker.html:
* fast/multicol/cell-shrinkback-expected.html:
* fast/multicol/cell-shrinkback.html:
* fast/multicol/newmulticol/breaks-3-columns-3-expected.html:
* fast/multicol/newmulticol/breaks-3-columns-3.html:
* fast/regions/last-region-border-radius-expected.html:
* fast/regions/last-region-border-radius.html:
* fast/regions/overflow/overflow-first-and-last-regions.html:
* fast/regions/overflow/overflow-in-uniform-regions-dynamic-expected.html:
* fast/regions/overflow/overflow-in-uniform-regions-dynamic.html:
* fast/shapes/shape-outside-floats/shape-outside-floats-inset-rounded-large-radius-expected.html:
* fast/shapes/shape-outside-floats/shape-outside-floats-inset-rounded-large-radius.html:
* fast/shapes/shape-outside-floats/shape-outside-rounded-inset-expected.html:
* fast/shapes/shape-outside-floats/shape-outside-rounded-inset.html:
* fast/table/td-width-fifty-percent-regression-expected.html:
* fast/table/td-width-fifty-percent-regression.html:
* fast/text/complex-initial-advance-expected.html:
* fast/text/complex-initial-advance.html:
* platform/mac/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180796 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Subpixel] Subpixelize RenderListMarker
mmaxfield@apple.com [Sat, 28 Feb 2015 00:23:15 +0000 (00:23 +0000)]
[Subpixel] Subpixelize RenderListMarker
https://bugs.webkit.org/show_bug.cgi?id=142093

Reviewed by Zalan Bujtas.

Source/WebCore:

Use floats instead of ints.

Test: fast/lists/rtl-marker.html

* rendering/RenderListMarker.cpp:
(WebCore::RenderListMarker::paint):
(WebCore::RenderListMarker::updateContent):
(WebCore::RenderListMarker::getRelativeMarkerRect):
* rendering/RenderListMarker.h:

LayoutTests:

* platform/mac/TestExpectations: Unskip fast/lists/rtl-marker.html

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180795 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago<attachment> should be selected immediately upon click, and be drag/copyable upon...
timothy_horton@apple.com [Sat, 28 Feb 2015 00:09:09 +0000 (00:09 +0000)]
<attachment> should be selected immediately upon click, and be drag/copyable upon click
https://bugs.webkit.org/show_bug.cgi?id=142114
<rdar://problem/19982520>

Reviewed by Enrica Casucci.

* css/html.css:
(attachment):
Make attachment use 'user-select: all' to act as a single click-to-select unit.

* html/HTMLAttachmentElement.cpp:
(WebCore::HTMLAttachmentElement::setFocus): Deleted.
* html/HTMLAttachmentElement.h:
* rendering/RenderAttachment.cpp:
(WebCore::RenderAttachment::isFocused): Deleted.
(WebCore::RenderAttachment::focusChanged): Deleted.
* rendering/RenderAttachment.h:
* rendering/RenderThemeMac.mm:
(WebCore::RenderThemeMac::paintAttachment):
Remove focus-related code; instead of focusing the element upon click,
we get a selection including just the <attachment>, and everything
behaves much more consistently (copy works, drag works, etc.).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180794 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoTest gardening, unmark now passing tests.
ap@apple.com [Sat, 28 Feb 2015 00:06:48 +0000 (00:06 +0000)]
Test gardening, unmark now passing tests.

* TestExpectations:
* platform/mac-wk2/TestExpectations:
* platform/mac/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180793 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[iOS] Some MathML tests crash in RenderMathMLOperator::advanceForGlyph() or boundsFor...
mmaxfield@apple.com [Sat, 28 Feb 2015 00:02:53 +0000 (00:02 +0000)]
[iOS] Some MathML tests crash in RenderMathMLOperator::advanceForGlyph() or boundsForGlyph()
https://bugs.webkit.org/show_bug.cgi?id=141371

Reviewed by David Kilzer.

Source/WebCore:

Null checks.

Covered by existing mathml tests.

* rendering/mathml/RenderMathMLOperator.cpp:
(WebCore::RenderMathMLOperator::boundsForGlyph):
(WebCore::RenderMathMLOperator::advanceForGlyph):
(WebCore::RenderMathMLOperator::getDisplayStyleLargeOperator):

LayoutTests:

Updating expected results and TestExpectations

* platform/ios-simulator-wk2/TestExpectations:
* platform/ios-simulator/mathml/opentype/horizontal-munderover-expected.txt:
* platform/ios-simulator/mathml/opentype/large-operators-expected.txt:
* platform/ios-simulator/mathml/opentype/vertical-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180792 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoMake SourceBuffer ActiveDOMObject suspendable
cdumez@apple.com [Fri, 27 Feb 2015 23:46:01 +0000 (23:46 +0000)]
Make SourceBuffer ActiveDOMObject suspendable
https://bugs.webkit.org/show_bug.cgi?id=142108
<rdar://problem/19923085>

Reviewed by Jer Noble.

Source/WebCore:

Make SourceBuffer ActiveDOMObject suspendable if it is removed from its
MediaSource and does not have any pending events. This makes it more
likely for pages using SourceBuffer objects to go into the PageCache.

Test: fast/history/page-cache-removed-source-buffer.html

* Modules/mediasource/SourceBuffer.cpp:
(WebCore::SourceBuffer::canSuspend):
* Modules/mediasource/SourceBuffer.h:

LayoutTests:

Add a layout tests to check that a SourceBuffer removed from its
MediaSource does not prevent a Page from entering PageCache.

* fast/history/page-cache-removed-source-buffer-expected.txt: Added.
* fast/history/page-cache-removed-source-buffer.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180791 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdd support for canvas ellipse method
dino@apple.com [Fri, 27 Feb 2015 23:40:35 +0000 (23:40 +0000)]
Add support for canvas ellipse method
https://bugs.webkit.org/show_bug.cgi?id=82791
<rdar://problem/11159172>

Patch by Sam Weinig <sam@webkit.org> on 2015-02-26
Reviewed by Dirk Schulze.

Source/WebCore:

Tests: fast/canvas/canvas-ellipse-360-winding.html
       fast/canvas/canvas-ellipse-circumference-fill.html
       fast/canvas/canvas-ellipse-circumference.html
       fast/canvas/canvas-ellipse-connecting-line.html
       fast/canvas/canvas-ellipse-negative-radius.html
       fast/canvas/canvas-ellipse-zero-lineto.html
       fast/canvas/canvas-ellipse.html

* html/canvas/CanvasPathMethods.h:
* html/canvas/CanvasPathMethods.cpp:
(WebCore::CanvasPathMethods::lineTo):
Convenience for passing a FloatPoint instead of two floats.

(WebCore::normalizeAngles):
Normalizes the angles as described in the HTML spec. Ensuring the startAngle
is greater than 0 and less than 2pi, and the the endAngle is at most 2pi
from the start angle.

(WebCore::CanvasPathMethods::arc):
- Renames some of the parameters to be clearer.
- Normalizes the angles for consistency with ellipse.
- Moves hasInvertibleTransform() higher in the function for consistency.

(WebCore::CanvasPathMethods::ellipse): Added.

* html/canvas/CanvasRenderingContext2D.idl:
* html/canvas/DOMPath.idl:
Add ellipse(...).

* platform/graphics/Path.h:
* platform/graphics/cg/PathCG.cpp:
(WebCore::Path::addArc):
Rename parameters for clarity and use a nullptr.

(WebCore::Path::addEllipse):
Added. Constructs an ellipse via a transformed arc.

LayoutTests:

* fast/canvas/canvas-ellipse-360-winding-expected.txt: Added.
* fast/canvas/canvas-ellipse-360-winding.html: Added.
* fast/canvas/canvas-ellipse-circumference-expected.txt: Added.
* fast/canvas/canvas-ellipse-circumference-fill-expected.txt: Added.
* fast/canvas/canvas-ellipse-circumference-fill.html: Added.
* fast/canvas/canvas-ellipse-circumference.html: Added.
* fast/canvas/canvas-ellipse-connecting-line-expected.html: Added.
* fast/canvas/canvas-ellipse-connecting-line.html: Added.
* fast/canvas/canvas-ellipse-expected.txt: Added.
* fast/canvas/canvas-ellipse-negative-radius-expected.txt: Added.
* fast/canvas/canvas-ellipse-negative-radius.html: Added.
* fast/canvas/canvas-ellipse-zero-lineto-expected.txt: Added.
* fast/canvas/canvas-ellipse-zero-lineto.html: Added.
* fast/canvas/canvas-ellipse.html: Added.
* fast/canvas/script-tests/canvas-ellipse-360-winding.js: Added.
* fast/canvas/script-tests/canvas-ellipse.js: Added.
* fast/canvas/script-tests/js-ellipse-implementation.js: Added.
* platform/mac/fast/canvas/canvas-ellipse-circumference-expected.png: Added.
* platform/mac/fast/canvas/canvas-ellipse-circumference-fill-expected.png: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180790 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoTest more features of content extensions.
commit-queue@webkit.org [Fri, 27 Feb 2015 23:37:07 +0000 (23:37 +0000)]
Test more features of content extensions.
https://bugs.webkit.org/show_bug.cgi?id=142100

Patch by Alex Christensen <achristensen@webkit.org> on 2015-02-27
Reviewed by Brady Eidson.

* http/tests/usercontentfilter/basic-filter.html:
* http/tests/usercontentfilter/basic-filter.html.json:
* platform/mac/http/tests/usercontentfilter/basic-filter-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180789 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWeb Inspector: Use Maps in ResourceCollection instead of objects
commit-queue@webkit.org [Fri, 27 Feb 2015 23:35:12 +0000 (23:35 +0000)]
Web Inspector: Use Maps in ResourceCollection instead of objects
https://bugs.webkit.org/show_bug.cgi?id=142101

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2015-02-27
Reviewed by Timothy Hatcher.

* UserInterface/Models/ResourceCollection.js:
(WebInspector.ResourceCollection):
(WebInspector.ResourceCollection.prototype.resourcesWithType):
(WebInspector.ResourceCollection.prototype.removeAllResources):
(WebInspector.ResourceCollection.prototype.resourceForURL):
(WebInspector.ResourceCollection.prototype._associateWithResource):
(WebInspector.ResourceCollection.prototype._disassociateWithResource):
(WebInspector.ResourceCollection.prototype._resourceURLDidChange):
(WebInspector.ResourceCollection.prototype._resourceTypeDidChange):
Use Maps instead of objects.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180788 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdd ability for run-jsc-benchmarks to set library path from test binary when run...
msaboff@apple.com [Fri, 27 Feb 2015 23:32:03 +0000 (23:32 +0000)]
Add ability for run-jsc-benchmarks to set library path from test binary when run on a build bot
https://bugs.webkit.org/show_bug.cgi?id=142112

Reviewed by Filip Pizlo.

Added check for VMs in the form of <someDir>/{DumpRenderTree,webkitTestRunner,jsc} and use
<someDir> as the library path.

* Scripts/run-jsc-benchmarks:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180787 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoSome WebGL tests fail on ATI hardware
ap@apple.com [Fri, 27 Feb 2015 23:31:57 +0000 (23:31 +0000)]
Some WebGL tests fail on ATI hardware
https://bugs.webkit.org/show_bug.cgi?id=93560
rdar://problem/19991477

* platform/mac/TestExpectations: Updating expectations, as this is not limited to
Mavericks.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180786 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdding support for serializing HTMLAttachment elements.
enrica@apple.com [Fri, 27 Feb 2015 23:07:03 +0000 (23:07 +0000)]
Adding support for serializing HTMLAttachment elements.
https://bugs.webkit.org/show_bug.cgi?id=142026

Reviewed by Tim Horton.

Source/WebCore:

Test: editing/pasteboard/copy-paste-attachment.html

Adding support to serialize the attachment element
and properly handle it when converting a DOM range
to NSAttributedString.

* editing/cocoa/HTMLConverter.mm:
(HTMLConverter::_processElement):
* editing/markup.cpp:
(WebCore::StyledMarkupAccumulator::appendCustomAttributes): Create new attribute
for attachment element when serializating.
(WebCore::StyledMarkupAccumulator::appendElement):
(WebCore::createFragmentFromMarkup): Remove the attribute from the attachment element
when creating the fragment.
* html/HTMLAttachmentElement.cpp:
(WebCore::HTMLAttachmentElement::file): Added const to file() to
use it in appendCustonAttributes where the element is a const reference.
* html/HTMLAttachmentElement.h:
* html/HTMLAttributeNames.in:

LayoutTests:

* editing/pasteboard/copy-paste-attachment-expected.txt: Added.
* editing/pasteboard/copy-paste-attachment.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180785 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago<attachment> should have an inactive style (gray in background)
timothy_horton@apple.com [Fri, 27 Feb 2015 23:00:03 +0000 (23:00 +0000)]
<attachment> should have an inactive style (gray in background)
https://bugs.webkit.org/show_bug.cgi?id=142103
<rdar://problem/19982486>

Reviewed by Dan Bernstein.

* rendering/RenderThemeMac.mm:
(WebCore::attachmentLabelInactiveBackgroundColor):
(WebCore::attachmentLabelInactiveTextColor):
(WebCore::RenderThemeMac::paintAttachmentLabelBackground):
(WebCore::RenderThemeMac::paintAttachmentLabel):
Use a gray background and gray text when the selection containing the
attachment isn't focused and active.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180784 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdd a "block-cookies" rule to the user content filter.
beidson@apple.com [Fri, 27 Feb 2015 22:50:03 +0000 (22:50 +0000)]
Add a "block-cookies" rule to the user content filter.
https://bugs.webkit.org/show_bug.cgi?id=142105

Reviewed by Alex Christensen.

Source/WebCore:

Tests: http/tests/usercontentfilter/block-cookies-basic.html
       http/tests/usercontentfilter/block-cookies-send.html

* contentextensions/ContentExtensionRule.h:

* contentextensions/ContentExtensionsBackend.cpp:
(WebCore::ContentExtensions::ContentExtensionsBackend::actionForURL):
(WebCore::ContentExtensions::ContentExtensionsBackend::shouldBlockURL): Deleted.
* contentextensions/ContentExtensionsBackend.h:

* contentextensions/ContentExtensionsManager.cpp:
(WebCore::ContentExtensions::ExtensionsManager::loadAction):

* loader/cache/CachedResourceLoader.cpp:
(WebCore::CachedResourceLoader::requestResource):

* page/UserContentController.cpp:
(WebCore::UserContentController::actionForURL):
(WebCore::UserContentController::contentFilterBlocksURL): Deleted.
* page/UserContentController.h:

LayoutTests:

* http/tests/cookies/resources/echo-cookies.php: Added.
* http/tests/usercontentfilter/block-cookies-basic-expected.txt: Added.
* http/tests/usercontentfilter/block-cookies-basic.html: Added.
* http/tests/usercontentfilter/block-cookies-basic.html.json: Added.
* http/tests/usercontentfilter/block-cookies-send-expected.txt: Added.
* http/tests/usercontentfilter/block-cookies-send.html: Added.
* http/tests/usercontentfilter/block-cookies-send.html.json: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180783 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoLookup panel dismisses when pages are loading in other tabs/windows
bdakin@apple.com [Fri, 27 Feb 2015 22:45:33 +0000 (22:45 +0000)]
Lookup panel dismisses when pages are loading in other tabs/windows
https://bugs.webkit.org/show_bug.cgi?id=142104
-and corresponding-
rdar://problem/19882137

Reviewed by Tim Horton.

Until rdar://problem/13875766 is resolved, we should only call into Lookup and
DataDetectors for key windows.
* UIProcess/API/mac/WKView.mm:
(-[WKView _dismissContentRelativeChildWindows]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180782 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[WinCairo] Unreviewed build fix.
achristensen@apple.com [Fri, 27 Feb 2015 22:17:09 +0000 (22:17 +0000)]
[WinCairo] Unreviewed build fix.

Source/WebCore:

* platform/graphics/BitmapImage.h:
* platform/win/BitmapInfo.h:
Added WEBCORE_EXPORT.

Tools:

* TestWebKitAPI/TestWebKitAPI.vcxproj/TestWebKitAPICommonWinCairo.props:
Use TestWebKitAPIPrefix.h like in TestWebKitAPICommon.props.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180781 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago<attachment>'s label baseline should match that of the surrounding text
timothy_horton@apple.com [Fri, 27 Feb 2015 21:56:43 +0000 (21:56 +0000)]
<attachment>'s label baseline should match that of the surrounding text
https://bugs.webkit.org/show_bug.cgi?id=142099
rdar://problem/19982495

Reviewed by Dan Bernstein.

* rendering/RenderAttachment.cpp:
(WebCore::RenderAttachment::baselinePosition):
* rendering/RenderAttachment.h:
Override baselinePosition and retrieve it from RenderTheme.

* rendering/RenderTheme.h:
(WebCore::RenderTheme::attachmentBaseline):
* rendering/RenderThemeMac.h:
* rendering/RenderThemeMac.mm:
(WebCore::AttachmentLayout::AttachmentLayout):
(WebCore::RenderThemeMac::attachmentBaseline):
Plumb the label baseline from AttachmentLayout to RenderAttachment.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180780 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed, rolling out r180752.
commit-queue@webkit.org [Fri, 27 Feb 2015 21:42:31 +0000 (21:42 +0000)]
Unreviewed, rolling out r180752.
https://bugs.webkit.org/show_bug.cgi?id=142098

Causes 10 SVG test failures on Windows. (Requested by
bfulgham_ on #webkit).

Reverted changeset:

"Cache glyph widths to GlyphPages"
https://bugs.webkit.org/show_bug.cgi?id=142028
http://trac.webkit.org/changeset/180752

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180779 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoiOS, GTK, and EFL rebaselines after r180726.
rniwa@webkit.org [Fri, 27 Feb 2015 21:40:05 +0000 (21:40 +0000)]
iOS, GTK, and EFL rebaselines after r180726.

* platform/efl/editing/inserting/5058163-1-expected.txt:
* platform/gtk/editing/inserting/5058163-1-expected.txt:
* platform/ios-simulator-wk2/editing/inserting/5058163-1-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180778 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[iOS] Fix build by defining EAGL_IOSURFACE macro before including <OpenGLES/EAGLPriva...
ddkilzer@apple.com [Fri, 27 Feb 2015 21:39:31 +0000 (21:39 +0000)]
[iOS] Fix build by defining EAGL_IOSURFACE macro before including <OpenGLES/EAGLPrivate.h>

* platform/spi/ios/OpenGLESSPI.h:  Define EAGL_IOSURFACE macro
until header refactoring is completed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180777 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed, skip tests added in r180771 on platforms that don't support IndexedDB.
cdumez@apple.com [Fri, 27 Feb 2015 21:34:51 +0000 (21:34 +0000)]
Unreviewed, skip tests added in r180771 on platforms that don't support IndexedDB.

* platform/efl/TestExpectations:
* platform/gtk/TestExpectations:
* platform/ios-simulator/TestExpectations:
* platform/win/TestExpectations:
* platform/wincairo/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180776 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Win] Skip IndexDB tests on Windows since it's not implemented.
bfulgham@apple.com [Fri, 27 Feb 2015 21:27:58 +0000 (21:27 +0000)]
[Win] Skip IndexDB tests on Windows since it's not implemented.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180775 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoMediaSource should be suspendable when closed
cdumez@apple.com [Fri, 27 Feb 2015 21:18:39 +0000 (21:18 +0000)]
MediaSource should be suspendable when closed
https://bugs.webkit.org/show_bug.cgi?id=142089
<rdar://problem/19923085>

Reviewed by Jer Noble.

Source/WebCore:

Make MediaSource ActiveDOMObject suspendable when it is in closed state
and it has no pending events. This increases the likelihood of pages
using MediaSource to enter the PageCache.

Tests: fast/history/page-cache-media-source-closed-2.html
       fast/history/page-cache-media-source-closed.html
       fast/history/page-cache-media-source-opened.html

LayoutTests:

Add layout tests to check that:
- Pages with an open MediaSource do not enter the PageCache
- Pages with an initially closed MediaSource enter the PageCache
- Pages with a MediaSource that changed state from opened to closed
  enter the PageCache.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180774 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdd infrastructure for handling website data in the network process
andersca@apple.com [Fri, 27 Feb 2015 20:14:22 +0000 (20:14 +0000)]
Add infrastructure for handling website data in the network process
https://bugs.webkit.org/show_bug.cgi?id=142092

Reviewed by Andreas Kling.

Source/WebKit2:

* NetworkProcess/NetworkProcess.cpp:
(WebKit::NetworkProcess::fetchWebsiteData):
(WebKit::NetworkProcess::deleteWebsiteDataForOrigins):
Send back "Did" messages without actually doing anything for now.

* NetworkProcess/NetworkProcess.h:
Add new members.

* NetworkProcess/NetworkProcess.messages.in:
Add FetchWebsiteData and DeleteWebsiteDataForOrigins messages.

* UIProcess/Network/NetworkProcessProxy.cpp:
(WebKit::NetworkProcessProxy::~NetworkProcessProxy):
Assert that all maps are empty.

(WebKit::NetworkProcessProxy::fetchWebsiteData):
(WebKit::NetworkProcessProxy::deleteWebsiteDataForOrigins):
Add callbacks and send fetch and delete messages respectively.

(WebKit::NetworkProcessProxy::networkProcessCrashedOrFailedToLaunch):
Make sure to invoke all callbacks.

(WebKit::NetworkProcessProxy::didFetchWebsiteData):
Find the callback and invoke it.

(WebKit::NetworkProcessProxy::didDeleteWebsiteDataForOrigins):
Ditto.

* UIProcess/Network/NetworkProcessProxy.h:
Add new members.

* UIProcess/Network/NetworkProcessProxy.messages.in:
Add DidFetchWebsiteData and DidDeleteWebsiteDataForOrigins messages.

* UIProcess/WebProcessProxy.cpp:
(WebKit::WebProcessProxy::connectionDidClose):
Just pass an empty WebsiteData object.

Tools:

Add a menu item that will fetch all website data, delete the returned data records, and
fetch all website data again so we can confirm that it's all empty.

* MiniBrowser/mac/WK2BrowserWindowController.m:
(-[WK2BrowserWindowController fetchAndClearWebsiteData:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180773 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUse NeverDestroyed for JS wrapper owners.
akling@apple.com [Fri, 27 Feb 2015 19:52:11 +0000 (19:52 +0000)]
Use NeverDestroyed for JS wrapper owners.
<https://webkit.org/b/142090>

Reviewed by Chris Dumez.

Using NeverDestroyed puts these objects in BSS which is preferable
since that prevents them from pinning down entire malloc pages forever.

* bindings/scripts/CodeGeneratorJS.pm:
(GenerateHeader): Use NeverDestroyed instead of DEPRECATED_DEFINE_STATIC_LOCAL.

* bindings/scripts/test/JS/*: Rebaseline bindings tests for this change.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180772 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoMake IDBDatabase / IDBRequest suspendable
cdumez@apple.com [Fri, 27 Feb 2015 19:24:25 +0000 (19:24 +0000)]
Make IDBDatabase / IDBRequest suspendable
https://bugs.webkit.org/show_bug.cgi?id=142076
<rdar://problem/19923085>

Reviewed by Andreas Kling.

Source/WebCore:

Make IDBDatabase / IDBRequest suspendable under certain conditions to
make it more likely for pages using indexeddb to enter the PageCache.

IDBDatabase is safely suspendable if the database is closed. IDBRequest
is safely suspendable if the request no longer has any pending activity
(i.e. state is DONE and success / failure handler was called). We may
be able to do better later but this is the bare minimum for now.

Tests: fast/history/page-cache-indexed-closed-db.html
       fast/history/page-cache-indexed-opened-db.html

* Modules/indexeddb/IDBDatabase.cpp:
(WebCore::IDBDatabase::IDBDatabase):
(WebCore::IDBDatabase::closeConnection):
(WebCore::IDBDatabase::enqueueEvent):
(WebCore::IDBDatabase::canSuspend):
* Modules/indexeddb/IDBDatabase.h:
* Modules/indexeddb/IDBRequest.cpp:
(WebCore::IDBRequest::canSuspend):
* Modules/indexeddb/IDBRequest.h:

LayoutTests:

Add layout tests to make sure that:
- A page with an open indexeddb database is not page-cacheable
- A page with a closed indexeddb database is page-cacheable

* fast/history/page-cache-indexed-closed-db-expected.txt: Added.
* fast/history/page-cache-indexed-closed-db.html: Added.
* fast/history/page-cache-indexed-opened-db-expected.txt: Added.
* fast/history/page-cache-indexed-opened-db.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180771 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoDrop unnecessary DatabaseManager::hasOpenDatabases() in PageCache::canCachePageContai...
cdumez@apple.com [Fri, 27 Feb 2015 19:22:10 +0000 (19:22 +0000)]
Drop unnecessary DatabaseManager::hasOpenDatabases() in PageCache::canCachePageContainingThisFrame()
https://bugs.webkit.org/show_bug.cgi?id=142052

Reviewed by Andreas Kling.

Source/WebCore:

Drop WebDatabase special-handling from PageCache::canCachePageContainingThisFrame().
DatabaseContext is already an ActiveDOMObject and DatabaseContext::canSuspend() was
returning false so pages using WebDatabase would never enter the PageCache anyway.

This patch also overrides ActiveDOMObject::canSuspend() in DatabaseContext to only
return false when there are open databases. This check is now equivalent to the one
that was in PageCache.

An issue that remains is that DatabaseContext::m_hasOpenDatabases is never reset
to false so once a page opened a database, it will never be page-cacheable. This
will be taken care of separately though.

Test: fast/history/page-cache-webdatabase-opened-db.html

* Modules/webdatabase/DatabaseContext.cpp:
(WebCore::DatabaseContext::canSuspend):
* Modules/webdatabase/DatabaseContext.h:
* history/PageCache.cpp:
(WebCore::logCanCacheFrameDecision):
(WebCore::PageCache::canCachePageContainingThisFrame):
* page/DiagnosticLoggingKeys.cpp:
(WebCore::DiagnosticLoggingKeys::hasOpenDatabasesKey): Deleted.
* page/DiagnosticLoggingKeys.h:

LayoutTests:

Add a layout test to check that a page with an open WebDatabase does
not enter the PageCache.

* fast/history/page-cache-webdatabase-opened-db-expected.txt: Added.
* fast/history/page-cache-webdatabase-opened-db.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180770 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoCompile DFA to bytecode.
achristensen@apple.com [Fri, 27 Feb 2015 19:20:27 +0000 (19:20 +0000)]
Compile DFA to bytecode.
https://bugs.webkit.org/show_bug.cgi?id=142031

Reviewed by Benjamin Poulain.

* WebCore.xcodeproj/project.pbxproj:
* contentextensions/ContentExtensionsBackend.cpp:
(WebCore::ContentExtensions::ContentExtensionsBackend::setRuleList):
(WebCore::ContentExtensions::ContentExtensionsBackend::shouldBlockURL):
* contentextensions/ContentExtensionsBackend.h:
* contentextensions/DFA.cpp:
(WebCore::ContentExtensions::DFA::nextState): Deleted.
(WebCore::ContentExtensions::DFA::actions): Deleted.
* contentextensions/DFA.h:
(WebCore::ContentExtensions::DFA::size):
(WebCore::ContentExtensions::DFA::nodeAt):
* contentextensions/DFABytecode.h: Added.
(WebCore::ContentExtensions::instructionSizeWithArguments):
* contentextensions/DFABytecodeCompiler.cpp: Added.
(WebCore::ContentExtensions::append):
(WebCore::ContentExtensions::set32Bits):
(WebCore::ContentExtensions::DFABytecodeCompiler::emitAppendAction):
(WebCore::ContentExtensions::DFABytecodeCompiler::emitJump):
(WebCore::ContentExtensions::DFABytecodeCompiler::emitCheckValue):
(WebCore::ContentExtensions::DFABytecodeCompiler::emitTerminate):
(WebCore::ContentExtensions::DFABytecodeCompiler::reserveBufferCapacity):
(WebCore::ContentExtensions::DFABytecodeCompiler::compileNode):
(WebCore::ContentExtensions::DFABytecodeCompiler::compile):
* contentextensions/DFABytecodeCompiler.h: Added.
(WebCore::ContentExtensions::DFABytecodeCompiler::DFABytecodeCompiler):
* contentextensions/DFABytecodeInterpreter.cpp: Added.
(WebCore::ContentExtensions::getBits):
(WebCore::ContentExtensions::DFABytecodeInterpreter::interpret):
* contentextensions/DFABytecodeInterpreter.h: Added.
(WebCore::ContentExtensions::DFABytecodeInterpreter::DFABytecodeInterpreter):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180769 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[WK2] REGRESSION(r180465): WebKit::WebPage::editorState() triggers a layout.
enrica@apple.com [Fri, 27 Feb 2015 18:50:20 +0000 (18:50 +0000)]
[WK2] REGRESSION(r180465): WebKit::WebPage::editorState() triggers a layout.
https://bugs.webkit.org/show_bug.cgi?id=142015

Reviewed by Alexey Proskuryakov.

We no longer compute the font information at selection
when we update the editor state.
Instead, we request the font information only when the selection
changes and the font panel is visible.
I added an observer to be notified of the font panel visibility
changes to update NSFontManager to reflect the font at the
current selection.

* Shared/EditorState.cpp:
(WebKit::EditorState::encode):
(WebKit::EditorState::decode):
* Shared/EditorState.h:
(WebKit::EditorState::EditorState):
* UIProcess/API/mac/WKView.mm:
(-[WKView updateFontPanelIfNeeded]):
(-[WKView _selectionChanged]):
(-[WKView addWindowObserversForWindow:]):
(-[WKView removeWindowObservers]):
(-[WKView observeValueForKeyPath:ofObject:change:context:]):
* UIProcess/WebPageProxy.h:
* UIProcess/WebPageProxy.messages.in:
* UIProcess/mac/WebPageProxyMac.mm:
(WebKit::WebPageProxy::fontAtSelection):
(WebKit::WebPageProxy::fontAtSelectionCallback):
* WebProcess/WebPage/WebPage.h:
* WebProcess/WebPage/WebPage.messages.in:
* WebProcess/WebPage/mac/WebPageMac.mm:
(WebKit::WebPage::platformEditorState):
(WebKit::WebPage::fontAtSelection):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180768 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUse after free in WebCore::RenderNamedFlowFragment::restoreRegionObjectsOriginalStyle
zalan@apple.com [Fri, 27 Feb 2015 18:43:48 +0000 (18:43 +0000)]
Use after free in WebCore::RenderNamedFlowFragment::restoreRegionObjectsOriginalStyle
https://bugs.webkit.org/show_bug.cgi?id=138366

Reviewed by Dave Hyatt.

This patch ensures that we clean up RenderNamedFlowFragment::m_renderObjectRegionStyle when embedded flow content is getting destroyed.

In m_renderObjectRegionStyle hash map, we store style information about the named flow's descendant children.
When a child is being detached from the tree, it removes itself from this hashmap.
We do it by traversing up on the ancestor chain and call removeFlowChildInfo() on the parent flow.
However in case of embedded flows (for example multicolumn content inside a region), we need to check whether the parent flow
is inside a flow too and continue the cleanup accordingly.

Source/WebCore:

Test: fast/regions/region-with-multicolumn-embedded-crash.html

* rendering/RenderObject.cpp:
(WebCore::RenderObject::removeFromRenderFlowThreadIncludingDescendants):

LayoutTests:

* fast/regions/region-with-multicolumn-embedded-crash-expected.txt: Added.
* fast/regions/region-with-multicolumn-embedded-crash.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180767 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdd API to remove a single content filter.
beidson@apple.com [Fri, 27 Feb 2015 18:33:39 +0000 (18:33 +0000)]
Add API to remove a single content filter.
<rdar://problem/19977764> and https://bugs.webkit.org/show_bug.cgi?id=142088

Reviewed by Sam Weinig.

Source/WebCore:

* page/UserContentController.cpp:
(WebCore::UserContentController::removeUserContentFilter):
* page/UserContentController.h:

Source/WebKit2:

* Shared/WebPageGroupData.h:

* UIProcess/API/C/WKPageGroup.cpp:
(WKPageGroupRemoveUserContentFilter):

* UIProcess/API/C/WKPageGroup.h:
* UIProcess/API/Cocoa/WKUserContentController.mm:
(-[WKUserContentController _removeUserContentFilter:]):

* UIProcess/API/Cocoa/WKUserContentControllerPrivate.h:
* UIProcess/UserContent/WebUserContentControllerProxy.cpp:
(WebKit::WebUserContentControllerProxy::WebUserContentControllerProxy):
(WebKit::WebUserContentControllerProxy::addProcess):
(WebKit::WebUserContentControllerProxy::addUserContentFilter):
(WebKit::WebUserContentControllerProxy::removeUserContentFilter):
(WebKit::WebUserContentControllerProxy::removeAllUserContentFilters):
* UIProcess/UserContent/WebUserContentControllerProxy.h:

* UIProcess/WebPageGroup.cpp:
(WebKit::WebPageGroup::addUserContentFilter):
(WebKit::WebPageGroup::removeUserContentFilter):
* UIProcess/WebPageGroup.h:

* WebProcess/UserContent/WebUserContentController.cpp:
(WebKit::WebUserContentController::removeUserContentFilter):
* WebProcess/UserContent/WebUserContentController.h:
* WebProcess/UserContent/WebUserContentController.messages.in:

* WebProcess/WebPage/WebPageGroupProxy.cpp:
(WebKit::WebPageGroupProxy::WebPageGroupProxy):
(WebKit::WebPageGroupProxy::removeUserContentFilter):
* WebProcess/WebPage/WebPageGroupProxy.h:
* WebProcess/WebPage/WebPageGroupProxy.messages.in:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180766 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Win] Remove remaining SafariTheme cruft
bfulgham@apple.com [Fri, 27 Feb 2015 18:09:46 +0000 (18:09 +0000)]
[Win] Remove remaining SafariTheme cruft
https://bugs.webkit.org/show_bug.cgi?id=142075

Reviewed by Anders Carlsson.

Remove reference to SafariTheme-switching preference.

Source/WebCore:

Tested by existing layout tests.

* WebCore.vcxproj/WebCore.vcxproj:
* WebCore.vcxproj/WebCore.vcxproj.filters:
* config.h:
* page/Settings.cpp:
(WebCore::Settings::fontRenderingMode):
(WebCore::Settings::setShouldPaintNativeControls): Deleted.
* page/Settings.h:
(WebCore::Settings::shouldPaintNativeControls): Deleted.
* platform/win/ScrollbarThemeWin.cpp:
(WebCore::ScrollbarTheme::nativeTheme):
* rendering/RenderThemeWin.cpp:
(WebCore::RenderTheme::themeForPage):

Source/WebKit/win:

* Interfaces/IWebPreferencesPrivate.idl:
* WebKitClassFactory.cpp:
(WebKitClassFactory::WebKitClassFactory):
(WebKitClassFactory::QueryInterface):
(WebKitClassFactory::AddRef):
(WebKitClassFactory::Release):
(WebKitClassFactory::CreateInstance):
(WebKitClassFactory::LockServer):
* WebPreferences.cpp:
(WebPreferences::shouldPaintNativeControls): Deleted.
(WebPreferences::setShouldPaintNativeControls): Deleted.
* WebPreferences.h:
* WebView.cpp:
(WebView::initWithFrame):
(WebView::notifyPreferencesChanged):

Tools:

* DumpRenderTree/win/DumpRenderTree.cpp:
(resetWebPreferencesToConsistentValues):
(prepareConsistentTestingEnvironment):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180765 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Win] Add another batch of debug assert failures.
bfulgham@apple.com [Fri, 27 Feb 2015 17:22:45 +0000 (17:22 +0000)]
[Win] Add another batch of debug assert failures.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180758 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Win] Rebaseline test after r180726.
bfulgham@apple.com [Fri, 27 Feb 2015 17:02:08 +0000 (17:02 +0000)]
[Win] Rebaseline test after r180726.

* platform/win/TestExpectations: Mark a flaky test.
* platform/win/editing/inserting/5058163-1-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180755 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdd comment about CSS value name mangling
mmaxfield@apple.com [Fri, 27 Feb 2015 16:59:18 +0000 (16:59 +0000)]
Add comment about CSS value name mangling

Unreviewed

* css/CSSParser.cpp:
(WebCore::cssValueKeywordID):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180754 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoCache glyph widths to GlyphPages
antti@apple.com [Fri, 27 Feb 2015 16:12:13 +0000 (16:12 +0000)]
Cache glyph widths to GlyphPages
https://bugs.webkit.org/show_bug.cgi?id=142028

Reviewed by Andreas Kling.

Currently we have a separate cache in Font for glyph widths. In practice we always need
the widths so we can just cache them in GlyphPages. This simplifies the code and removes
a per-character hash lookup from WidthIterator.

* platform/graphics/Font.cpp:
(WebCore::Font::Font):
(WebCore::Font::initCharWidths):
(WebCore::Font::platformGlyphInit):
(WebCore::createAndFillGlyphPage):
(WebCore::Font::computeWidthForGlyph):

    Rename to make it clear this doesn't cache.

(WebCore::GlyphPage::setGlyphDataForIndex):

    Initialize the width.
    This could go to GlyphPage.cpp if we had one.

* platform/graphics/Font.h:
(WebCore::Font::glyphZeroWidth):
(WebCore::Font::isZeroWidthSpaceGlyph):
(WebCore::Font::zeroGlyph): Deleted.
(WebCore::Font::setZeroGlyph): Deleted.
(WebCore::Font::widthForGlyph): Deleted.
* platform/graphics/FontCascade.cpp:
(WebCore::offsetToMiddleOfGlyph):
* platform/graphics/FontCascadeFonts.cpp:
(WebCore::FontCascadeFonts::glyphDataForCharacter):
* platform/graphics/GlyphPage.h:
(WebCore::GlyphData::GlyphData):

    Return width too as part of GlyphData.

(WebCore::GlyphPage::glyphDataForIndex):
(WebCore::GlyphPage::setGlyphDataForCharacter):
(WebCore::GlyphPage::setGlyphDataForIndex):
(WebCore::GlyphPage::GlyphPage):
* platform/graphics/WidthIterator.cpp:
(WebCore::WidthIterator::advanceInternal):

    No need to lookup width separately now.

* platform/graphics/mac/ComplexTextController.cpp:
(WebCore::ComplexTextController::adjustGlyphsAndAdvances):
* platform/graphics/mac/ComplexTextControllerCoreText.mm:
(WebCore::ComplexTextController::ComplexTextRun::ComplexTextRun):
* rendering/mathml/RenderMathMLOperator.cpp:
(WebCore::RenderMathMLOperator::advanceForGlyph):
* rendering/svg/SVGTextRunRenderingContext.cpp:
(WebCore::missingGlyphForFont):
* svg/SVGFontData.cpp:
(WebCore::SVGFontData::initializeFont):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180752 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoUnreviewed, roll out r180723. It broke a bunch of tests.
fpizlo@apple.com [Fri, 27 Feb 2015 07:26:23 +0000 (07:26 +0000)]
Unreviewed, roll out r180723. It broke a bunch of tests.

* bytecompiler/BytecodeGenerator.cpp:
(JSC::BytecodeGenerator::constLocal):
* bytecompiler/BytecodeGenerator.h:
* bytecompiler/NodesCodegen.cpp:
(JSC::ConstDeclNode::emitCodeSingle):
* tests/stress/const-arguments.js: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180732 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoiOS build fix after r180717.
ap@apple.com [Fri, 27 Feb 2015 06:06:46 +0000 (06:06 +0000)]
iOS build fix after r180717.

* Modules/webdatabase/DatabaseThread.cpp: (WebCore::DatabaseUnpauseTask::doPerformTask):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180727 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoisEditablePosition and related functions shouldn't move position out of table
rniwa@webkit.org [Fri, 27 Feb 2015 05:51:31 +0000 (05:51 +0000)]
isEditablePosition and related functions shouldn't move position out of table
https://bugs.webkit.org/show_bug.cgi?id=129200

Reviewed by Darin Adler.

Source/WebCore:

This patch removes the legacy editing position for elements display: table in its computed style.
Previously, we used (table, 0) and (table, !0) to denote positions immediately before and after
such an element for historical reasons. This forced us to update the style tree before computing
the editability of a position because we have to check the editability of the position outside
the element with display: table if the position was using such a legacy editing position.
e.g. if a table was not editable (contenteditable=false), the position before the table (table, 0)
should still be considered editable if the parent node of the table was editable.

This patch replaces such a legacy editing position by using modern position types:
PositionIsBeforeAnchor and PositionIsAfterAnchor.

No new tests since there should be no change in the user perceived editing operations.

* dom/Position.cpp:
(WebCore::Position::previous): Setup the node and the offset correctly when the original position's
type is PositionIsBeforeAnchor. Also return a position before or after node when the node we found
is "atomic" (e.g. input, img, br, etc...) or it's a table. This avoids creating a legacy editing
position inside a table.
(WebCore::Position::next): Ditto.
(WebCore::Position::atStartOfTree): Use atFirstEditingPositionForNode, which takes care of all types
of positions.
(WebCore::Position::atEndOfTree): Ditto.
(WebCore::Position::downstream): Return a position before a node instead of a legacy editing position
for an atomic element or a table element as done in the equivalent code in Position::upstream.
(WebCore::Position::isCandidate): Don't treat a position inside a table to be a candidate. e.g.
(table, 1) when there are more than two children of the table.

* dom/PositionIterator.cpp:
(WebCore::PositionIterator::operator Position): PositionIterator internally uses legacy editing
positions. So convert it to a modern position by returning a position before or after a table here.
* editing/ApplyBlockElementCommand.cpp:
(WebCore::ApplyBlockElementCommand::formatSelection): Check that the unsplittable element we found
is actually empty before executing the simple code path for an empty unsplittable element. Without
this check, block formatting a table element will fail.

* editing/htmlediting.cpp:
(WebCore::isEditablePosition): Use containerNode instead of deprecatedNode because the editability
of a position before or after an element is determined by its parent, not the element itself.
(WebCore::isAtUnsplittableElement): Ditto.
(WebCore::isRichlyEditablePosition): Ditto. Removed the code that moved the starting node out of
an element with display: table. This is the code removal for which this patch was made.
(WebCore::editableRootForPosition): Ditto.

LayoutTests:

Rebaselined a test. There is no visual difference.

* platform/mac/editing/inserting/5058163-1-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180726 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWeb Inspector: Add a subtle blue background for selected console messages
nvasilyev@apple.com [Fri, 27 Feb 2015 05:32:07 +0000 (05:32 +0000)]
Web Inspector: Add a subtle blue background for selected console messages
https://bugs.webkit.org/show_bug.cgi?id=142073

Reviewed by Timothy Hatcher.

* UserInterface/Views/LogContentView.css:
(.console-item.selected::after):
(.console-messages:focus .console-item.selected):
(.console-messages:focus .console-item.selected + .console-item):
(.console-error-level + .console-item):
(.console-error-level .section .header .title):
(.console-warning-level):
(.console-warning-level + .console-item):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180725 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Win] More test expectation updates.
bfulgham@apple.com [Fri, 27 Feb 2015 05:27:13 +0000 (05:27 +0000)]
[Win] More test expectation updates.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180724 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoBytecodeGenerator::constLocal() behaves identically to BytecodeGenerator::local(...
fpizlo@apple.com [Fri, 27 Feb 2015 05:12:46 +0000 (05:12 +0000)]
BytecodeGenerator::constLocal() behaves identically to BytecodeGenerator::local() for the purposes of its one caller
https://bugs.webkit.org/show_bug.cgi?id=142071

Rubber stamped by Benjamin Poulain.

The only behavioral differences between constLocal() and local() are:

- constLocal() doesn't have a special case for "this" that overrides other checks like the
  shouldOptimizeLocals() check.  But the one user of constLocal() is for the "const x"
  expression, and "const this" doesn't parse.

- constLocal() won't createArgumentsIfNecessary() for "arguments".  But it's harmless if
  it does, since its one user assigns to the local.

So, we can remove constLocal() and make its one caller use local() instead.

* bytecompiler/BytecodeGenerator.cpp:
(JSC::BytecodeGenerator::constLocal): Deleted.
* bytecompiler/BytecodeGenerator.h:
* bytecompiler/NodesCodegen.cpp:
(JSC::ConstDeclNode::emitCodeSingle):
* tests/stress/const-arguments.js: Added.
(foo):
(check):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180723 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWeb Inspector: Colored console messages apply text color to object tree properties
commit-queue@webkit.org [Fri, 27 Feb 2015 04:11:03 +0000 (04:11 +0000)]
Web Inspector: Colored console messages apply text color to object tree properties
https://bugs.webkit.org/show_bug.cgi?id=142051

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2015-02-26
Reviewed by Timothy Hatcher.

* UserInterface/Views/ObjectPreviewView.css:
(.object-preview):
* UserInterface/Views/ObjectTreeView.css:
(.object-tree):
Default the color to black within Object Tree and Object Preview views.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180722 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWeb Inspector: Clear Log on Reload clears some logs that were after reload
commit-queue@webkit.org [Fri, 27 Feb 2015 04:08:24 +0000 (04:08 +0000)]
Web Inspector: Clear Log on Reload clears some logs that were after reload
https://bugs.webkit.org/show_bug.cgi?id=142070

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2015-02-26
Reviewed by Timothy Hatcher.

* UserInterface/Views/LogContentView.js:
(WebInspector.LogContentView.prototype._sessionStarted):
We can bail after calling clearLog, since that will already start
a new session for us.

(WebInspector.LogContentView.prototype._clearLog):
Don't trigger a backend clear messages, since that will cause us to
clear messages received between the request and response.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180721 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoImplement <attachment> element appearance on Mac
timothy_horton@apple.com [Fri, 27 Feb 2015 02:07:22 +0000 (02:07 +0000)]
Implement <attachment> element appearance on Mac
https://bugs.webkit.org/show_bug.cgi?id=142023

Reviewed by Dean Jackson.

* css/CSSParser.cpp:
(WebCore::isValidKeywordPropertyAndValue):
* css/CSSPrimitiveValueMappings.h:
(WebCore::CSSPrimitiveValue::CSSPrimitiveValue):
* css/CSSValueKeywords.in:
* css/html.css:
(attachment):
* platform/ThemeTypes.h:
Add a new -webkit-appearance value, attachment.

* WebCore.xcodeproj/project.pbxproj:
* html/HTMLAttachmentElement.cpp:
(WebCore::HTMLAttachmentElement::setFile):
Let the renderer know when the file changes (so it can relayout and repaint).

(WebCore::HTMLAttachmentElement::filePath):
(WebCore::HTMLAttachmentElement::fileName):
* html/HTMLAttachmentElement.h:
Handy acccessors for the full path and the name of the file.

* platform/spi/mac/IconServicesSPI.h: Added.
* platform/spi/mac/LaunchServicesSPI.h: Added.
Add some IconServices and LaunchServices SPI headers.

* rendering/RenderAttachment.cpp:
(WebCore::RenderAttachment::RenderAttachment):
Don't have a default intrinsic size; we'll adjust in layout().

(WebCore::RenderAttachment::isSelected):
(WebCore::RenderAttachment::isFocused):
(WebCore::RenderAttachment::isSelectedOrFocused):
Factor out from paintReplaced().

(WebCore::RenderAttachment::layout):
Use RenderTheme to compute the intrinsic size of this attachment.

(WebCore::RenderAttachment::paintReplaced):
Use RenderTheme to paint the attachment.

(WebCore::RenderAttachment::representedFileChanged):
When the represented file changes, we need to recompute the intrinsic size and repaint.

* rendering/RenderTheme.cpp:
(WebCore::RenderTheme::adjustStyle):
(WebCore::RenderTheme::paint):
(WebCore::RenderTheme::adjustAttachmentStyle):
(WebCore::RenderTheme::paintAttachment):
Plumb through the attachment appearance.

* rendering/RenderAttachment.h:
* rendering/RenderTheme.h:
(WebCore::RenderTheme::paintAttachment):
(WebCore::RenderTheme::attachmentIntrinsicSize):
* rendering/RenderThemeMac.h:
* rendering/RenderThemeMac.mm:
(WebCore::attachmentIconBackgroundColor):
(WebCore::attachmentIconBorderColor):
(WebCore::AttachmentLayout):
AttachmentLayout lays out the innards of the RenderThemeMac attachment
presentation: an icon with an optional bordered background, with a label
underneath it with an optional background.

(WebCore::RenderThemeMac::attachmentIntrinsicSize):
(WebCore::paintAttachmentIconBackground):
(WebCore::paintAttachmentIcon):
(WebCore::RenderThemeMac::paintAttachmentLabelBackground):
(WebCore::RenderThemeMac::paintAttachmentLabel):
(WebCore::RenderThemeMac::paintAttachment):
Paint the attachment element. Paint the optional backgrounds (and swap
out the text color) if the element is selected/focused.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180720 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAssertion fix for r180711: The bool returning form of BytecodeGenerator::addVar(...
mark.lam@apple.com [Fri, 27 Feb 2015 02:07:02 +0000 (02:07 +0000)]
Assertion fix for r180711: The bool returning form of BytecodeGenerator::addVar() can be removed.
<https://webkit.org/b/142064>

Reviewed by Joseph Pecoraro.

* bytecompiler/BytecodeGenerator.cpp:
(JSC::BytecodeGenerator::addVar):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180719 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAX: Expose caret browsing preference to accessibility API
commit-queue@webkit.org [Fri, 27 Feb 2015 01:57:48 +0000 (01:57 +0000)]
AX: Expose caret browsing preference to accessibility API
https://bugs.webkit.org/show_bug.cgi?id=141862

Patch by Doug Russell <d_russell@apple.com> on 2015-02-26
Reviewed by Chris Fleizach.

Exposing the caret browsing setting on WebCore::Frame via the accessibility API would allow assistive tech apps to enable it contextually (for example, when the assistive tech app is running).
Enabling caret browsing when assistive tech apps are running greatly improves the reliability of keyboard navigation on the web. Most especially in cases where selection would be disrupted by focus events.

Source/WebCore:

Tests: platform/mac/accessibility/caret-browsing-arrow-nav.html
       platform/mac/accessibility/caret-browsing-attribute.html
       platform/mac/accessibility/caret-browsing-tab-selection.html

* accessibility/AccessibilityObject.h:
* accessibility/mac/AccessibilityObjectMac.mm:
(WebCore::AccessibilityObject::caretBrowsingEnabled):
(WebCore::AccessibilityObject::setCaretBrowsingEnabled):
* accessibility/mac/WebAccessibilityObjectWrapperMac.mm:
(-[WebAccessibilityObjectWrapper accessibilityAttributeNames]):
(-[WebAccessibilityObjectWrapper accessibilityAttributeValue:]):
(-[WebAccessibilityObjectWrapper accessibilityIsAttributeSettable:]):
(-[WebAccessibilityObjectWrapper _accessibilitySetValue:forAttribute:]):
(-[WebAccessibilityObjectWrapper _accessibilitySetTestValue:forAttribute:]):

Tools:

* DumpRenderTree/AccessibilityUIElement.cpp:
(setBoolAttributeValueCallback):
(selectedTextMarkerRangeCallback):
(resetSelectedTextMarkerRangeCallback):
(AccessibilityUIElement::setBoolAttributeValue):
(AccessibilityUIElement::selectedTextMarkerRange):
(AccessibilityUIElement::resetSelectedTextMarkerRange):
(AccessibilityUIElement::getJSClass):
* DumpRenderTree/AccessibilityUIElement.h:
* DumpRenderTree/mac/AccessibilityUIElementMac.mm:
(AccessibilityUIElement::setBoolAttributeValue):
(AccessibilityUIElement::selectedTextMarkerRange):
(AccessibilityUIElement::resetSelectedTextMarkerRange):
* WebKitTestRunner/InjectedBundle/AccessibilityUIElement.cpp:
(WTR::AccessibilityUIElement::selectedTextMarkerRange):
(WTR::AccessibilityUIElement::resetSelectedTextMarkerRange):
(WTR::AccessibilityUIElement::setBoolAttributeValue):
* WebKitTestRunner/InjectedBundle/AccessibilityUIElement.h:
* WebKitTestRunner/InjectedBundle/Bindings/AccessibilityUIElement.idl:
* WebKitTestRunner/InjectedBundle/mac/AccessibilityUIElementMac.mm:
(WTR::AccessibilityUIElement::setBoolAttributeValue):
(WTR::AccessibilityUIElement::selectedTextMarkerRange):
(WTR::AccessibilityUIElement::resetSelectedTextMarkerRange):

LayoutTests:

* accessibility/parent-delete-expected.txt:
* platform/mac/accessibility/caret-browsing-arrow-nav-expected.txt: Added.
* platform/mac/accessibility/caret-browsing-arrow-nav.html: Added.
* platform/mac/accessibility/caret-browsing-attribute-expected.txt: Added.
* platform/mac/accessibility/caret-browsing-attribute.html: Added.
* platform/mac/accessibility/caret-browsing-tab-selection-expected.txt: Added.
* platform/mac/accessibility/caret-browsing-tab-selection.html: Added.
* platform/mac/accessibility/document-attributes-expected.txt:
* platform/mac/accessibility/resources/accessibility-helper.js: Added.
(clearSelectionAndFocusOnWebArea):
(elementAtStartMarkerOfSelectedTextMarkerRange):
(caretBrowsingEnabled):
(setCaretBrowsingEnabled):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180718 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoRemove unnecessary create() factory functions
gyuyoung.kim@samsung.com [Fri, 27 Feb 2015 01:56:38 +0000 (01:56 +0000)]
Remove unnecessary create() factory functions
https://bugs.webkit.org/show_bug.cgi?id=142039

Reviewed by Chris Dumez.

Clean up remaining JSCryptoKeySerializationJWK::create, HTMLMediaSession::create,
and DatabaseUnpauseTask::create.

* Modules/webdatabase/DatabaseThread.cpp:
(WebCore::DatabaseUnpauseTask::DatabaseUnpauseTask):
(WebCore::DatabaseThread::setPaused):
(WebCore::DatabaseUnpauseTask::create): Deleted.
* bindings/js/JSCryptoKeySerializationJWK.h:
* bindings/js/JSSubtleCryptoCustom.cpp:
(WebCore::importKey):
* html/HTMLMediaElement.cpp:
(WebCore::HTMLMediaElement::HTMLMediaElement):
* html/HTMLMediaSession.cpp:
(WebCore::HTMLMediaSession::create): Deleted.
* html/HTMLMediaSession.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180717 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoMachineThreads::Thread clean up has a use after free race condition.
mark.lam@apple.com [Fri, 27 Feb 2015 01:25:21 +0000 (01:25 +0000)]
MachineThreads::Thread clean up has a use after free race condition.
<https://webkit.org/b/141990>

Reviewed by Filip Pizlo.

MachineThreads::Thread clean up relies on the clean up mechanism
implemented in _pthread_tsd_cleanup_key(), which looks like this:

void _pthread_tsd_cleanup_key(pthread_t self, pthread_key_t key)
{
    void (*destructor)(void *);
    if (_pthread_key_get_destructor(key, &destructor)) {
        void **ptr = &self->tsd[key];
        void *value = *ptr;

    // === Start of window for the bug to manifest =================

        // At this point, this thread has cached "destructor" and "value"
        // (which is a MachineThreads*).  If the VM gets destructed (along
        // with its MachineThreads registry) by another thread, then this
        // thread will have no way of knowing that the MachineThreads* is
        // now pointing to freed memory.  Calling the destructor below will
        // therefore result in a use after free scenario when it tries to
        // access the MachineThreads' data members.

        if (value) {
            *ptr = NULL;
            if (destructor) {

    // === End of window for the bug to manifest ==================

                destructor(value);
            }
        }
    }
}

The fix is to add each active MachineThreads to an ActiveMachineThreadsManager,
and always check if the manager still contains that MachineThreads object
before we call removeCurrentThread() on it.  When MachineThreads is destructed,
it will remove itself from the manager.  The add, remove, and checking
operations are all synchronized on the manager's lock, thereby ensuring that
the MachineThreads object, if found in the manager, will remain alive for the
duration of time we call removeCurrentThread() on it.

There's also possible for the MachineThreads object to already be destructed
and another one happened to have been instantiated at the same address.
Hence, we should only remove the exiting thread if it is found in the
MachineThreads object.

There is no test for this issue because this bug requires a race condition
between 2 threads where:
1. Thread B, which had previously used the VM, exiting and
   getting to the bug window shown in _pthread_tsd_cleanup_key() above.
2. Thread A destructing the VM (and its MachineThreads object)
   within that window of time before Thread B calls the destructor.

It is not possible to get a reliable test case without invasively
instrumenting _pthread_tsd_cleanup_key() or MachineThreads::removeCurrentThread()
to significantly increase that window of opportunity.

* heap/MachineStackMarker.cpp:
(JSC::ActiveMachineThreadsManager::Locker::Locker):
(JSC::ActiveMachineThreadsManager::add):
(JSC::ActiveMachineThreadsManager::remove):
(JSC::ActiveMachineThreadsManager::contains):
(JSC::ActiveMachineThreadsManager::ActiveMachineThreadsManager):
(JSC::activeMachineThreadsManager):
(JSC::MachineThreads::MachineThreads):
(JSC::MachineThreads::~MachineThreads):
(JSC::MachineThreads::removeThread):
(JSC::MachineThreads::removeThreadIfFound):
(JSC::MachineThreads::removeCurrentThread): Deleted.
* heap/MachineStackMarker.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180716 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWeb Inspector: Save Console Evaluations into Command Line variables $1-$99 ($n)
joepeck@webkit.org [Fri, 27 Feb 2015 01:15:23 +0000 (01:15 +0000)]
Web Inspector: Save Console Evaluations into Command Line variables $1-$99 ($n)
https://bugs.webkit.org/show_bug.cgi?id=142061

Reviewed by Timothy Hatcher.

Source/JavaScriptCore:

* inspector/protocol/Debugger.json:
* inspector/protocol/Runtime.json:
Input flag "saveResult" on whether we should try to save a result.
Output int "savedResultIndex" to tell the frontend the saved state.

* inspector/InjectedScriptSource.js:
Handle saving and clearing $1-$99 values.
Include in BasicCommandLineAPI for JSContext inspection.

* inspector/InjectedScriptBase.cpp:
(Inspector::InjectedScriptBase::makeEvalCall):
* inspector/InjectedScriptBase.h:
Allow an optional "savedResultIndex" out value on evals.

* inspector/InjectedScript.cpp:
(Inspector::InjectedScript::evaluate):
(Inspector::InjectedScript::evaluateOnCallFrame):
* inspector/InjectedScript.h:
* inspector/agents/InspectorDebuggerAgent.cpp:
(Inspector::InspectorDebuggerAgent::evaluateOnCallFrame):
* inspector/agents/InspectorDebuggerAgent.h:
* inspector/agents/InspectorRuntimeAgent.cpp:
(Inspector::InspectorRuntimeAgent::evaluate):
* inspector/agents/InspectorRuntimeAgent.h:
Plumbing for new in and out parameters.

Source/WebCore:

* inspector/CommandLineAPIModuleSource.js:
Replace $1-$4 "inspected objects" with $1-$99 "saved results".

* bindings/js/JSCommandLineAPIHostCustom.cpp:
(WebCore::JSCommandLineAPIHost::inspectedObject):
* inspector/CommandLineAPIHost.cpp:
(WebCore::CommandLineAPIHost::CommandLineAPIHost):
(WebCore::CommandLineAPIHost::InspectableObject::get):
(WebCore::CommandLineAPIHost::addInspectedObject):
(WebCore::CommandLineAPIHost::inspectedObject):
(WebCore::CommandLineAPIHost::clearInspectedObjects): Deleted.
* inspector/CommandLineAPIHost.h:
* inspector/CommandLineAPIHost.idl:
Since we now just save the single $0 inspected object, eliminate
keeping track of a list of 5 values.

Source/WebInspectorUI:

* UserInterface/Controllers/RuntimeManager.js:
(WebInspector.RuntimeManager.prototype.evalCallback):
(WebInspector.RuntimeManager.prototype.evaluateInInspectedWindow):
Add a saveResult parameter for the new protocol command in parameter.

* UserInterface/Views/ConsoleMessageImpl.js:
(WebInspector.ConsoleMessageImpl.prototype._formatMessage):
(WebInspector.ConsoleMessageImpl.prototype._formatParameterAsObject):
(WebInspector.ConsoleMessageImpl.prototype._formatParameterAsArray):
(WebInspector.ConsoleMessageImpl.prototype._rootPropertyPathForObject):
* UserInterface/Views/ObjectTreeView.js:
(WebInspector.ObjectTreeView.prototype.appendTitleSuffix):
For console evaluation results, show a "= $n" when the evaluation was
given a saved result index.

* UserInterface/Views/LogContentView.css:
(.console-saved-variable):
* UserInterface/Views/ObjectPreviewView.css:
(.object-preview-name):
Make $n and class names in previews always non-italics.

* UserInterface/Controllers/JavaScriptLogViewController.js:
(WebInspector.JavaScriptLogViewController.prototype.printResult):
(WebInspector.JavaScriptLogViewController.prototype.consolePromptTextCommitted):
* UserInterface/Controllers/JavaScriptRuntimeCompletionProvider.js:
(get WebInspector.JavaScriptRuntimeCompletionProvider.prototype.):
* UserInterface/Controllers/LogManager.js:
(WebInspector.LogManager.prototype.messagesCleared):
(WebInspector.LogManager.prototype._mainResourceDidChange):
Try to provide better autocompletion for $n, by populating autocompletion menus
from $1-$n where n is the maximum saved result index seen. Clear the maximum
when we clear the console.

* UserInterface/Views/ConsoleCommandResult.js:
(WebInspector.ConsoleCommandResult):
(WebInspector.ConsoleCommandResult.clearMaximumSavedResultIndex):
Keep track of the maximum savedResultIndex for console evaluation results.

LayoutTests:

* inspector/debugger/command-line-api-exception-nested-catch.html:
* inspector/debugger/command-line-api-exception.html:
* inspector/model/remote-object-get-properties.html:
* inspector/model/remote-object-weak-collection.html:
* inspector/model/remote-object.html:
Update evaluateInInspectedWindow call sites for new parameter.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180715 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWeb Inspector: Set/Map appear as lossless when they have lossy entries
joepeck@webkit.org [Fri, 27 Feb 2015 01:15:17 +0000 (01:15 +0000)]
Web Inspector: Set/Map appear as lossless when they have lossy entries
https://bugs.webkit.org/show_bug.cgi?id=142050

Reviewed by Timothy Hatcher.

* UserInterface/Views/ObjectPreviewView.js:
(WebInspector.ObjectPreviewView.prototype._appendEntryPreviews):
Take into account the lossless states of entry keys and values.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180714 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoWeb Inspector: New ObjectTree UI for Arrays / Maps / Sets
joepeck@webkit.org [Fri, 27 Feb 2015 01:15:15 +0000 (01:15 +0000)]
Web Inspector: New ObjectTree UI for Arrays / Maps / Sets
https://bugs.webkit.org/show_bug.cgi?id=142037

Reviewed by Timothy Hatcher.

* Localizations/en.lproj/localizedStrings.js:
* UserInterface/Main.html:
Miscellaneous changes.

* UserInterface/Models/PropertyDescriptor.js:
(WebInspector.PropertyDescriptor.prototype.isIndexProperty):
Useful for quickly checking if this property is numeric and possibly
an array index.

* UserInterface/Models/PropertyPath.js:
(WebInspector.PropertyPath.prototype.appendMapKey):
(WebInspector.PropertyPath.prototype.appendMapValue):
(WebInspector.PropertyPath.prototype.appendSetIndex):
Be specific about property paths into maps / sets. Note that a map
value may be displayable if the key is simple.

* UserInterface/Protocol/RemoteObject.js:
(WebInspector.RemoteObject.prototype.hasValue):
A simple value RemoteObject may have the value "undefined". So provide
a falsey proof helper that actually checks if we have a value.

(WebInspector.RemoteObject.prototype.isArray):
(WebInspector.RemoteObject.prototype.backendGetOwnPropertyDescriptor):
(WebInspector.RemoteObject.prototype.wrappedCallback):
(WebInspector.RemoteObject.prototype.getOwnPropertyDescriptor):
Currently backend APIs exist only to get all properties. In the case
of collections, we often want to get only one property (__proto__).
This is a simple implementation on top of callFunctionOn.

* UserInterface/Views/ConsoleMessageImpl.js:
(WebInspector.ConsoleMessageImpl.prototype._formatParameterAsObject):
(WebInspector.ConsoleMessageImpl.prototype._formatParameterAsArray):
(WebInspector.ConsoleMessageImpl.prototype.appendUndefined): Deleted.
(WebInspector.ConsoleMessageImpl.prototype._printArray): Deleted.
(WebInspector.ConsoleMessageImpl.prototype._formatAsArrayEntry): Deleted.
Simplify array formatted to just use an ObjectTreeView. Add fixmes
that we should seed the ObjectTreeView with a starting property path.

* UserInterface/Views/FormattedValue.css:
(.formatted-node > ol):
Sometimes, a node's display was getting overridden by various console styles.
Force a node to always display block. We may be able to remove this later.

* UserInterface/Views/FormattedValue.js:
(WebInspector.FormattedValue.createObjectTreeOrFormattedValueForRemoteObject):
Helper for formatting a node / object / value more easily. This
is used by all collection types.

* UserInterface/Views/ObjectPreviewView.js:
(WebInspector.ObjectPreviewView.prototype._appendPropertyPreviews):
We lost the nice sparse array support when switching to the new preview path,
we should add it back.

* UserInterface/Views/ObjectTreeArrayIndexTreeElement.css:
(.object-tree-array-index):
(.object-tree-array-index > .titles):
(.object-tree-array-index > .icon):
(.object-tree-array-index .index-name):
(.object-tree-array-index .index-value .object-tree):
(.object-tree-array-index .index-value .object-tree .object-tree-outline):
(.object-tree-property + ol .object-tree-array-index):
New styles specific to array index tree elements.

* UserInterface/Views/ObjectTreeMapEntryTreeElement.css:
(.object-tree-array-index.object-tree-map-entry > .titles > .title > .index-name):
(.object-tree-map-entry.key):
(.object-tree-map-entry.key:first-of-type):
(.object-tree-map-entry):
New styles specific to map key/value tree elements.

* UserInterface/Views/ObjectTreeCollectionTreeElement.js: Removed.
Remove old collection implementation.

* UserInterface/Views/ObjectTreeArrayIndexTreeElement.js: Added.
(WebInspector.ObjectTreeArrayIndexTreeElement):
(WebInspector.ObjectTreeArrayIndexTreeElement.prototype.get property):
(WebInspector.ObjectTreeArrayIndexTreeElement.prototype._resolvedValue):
(WebInspector.ObjectTreeArrayIndexTreeElement.prototype._propertyPathType):
(WebInspector.ObjectTreeArrayIndexTreeElement.prototype._resolvedValuePropertyPath):
(WebInspector.ObjectTreeArrayIndexTreeElement.prototype._thisPropertyPath):
(WebInspector.ObjectTreeArrayIndexTreeElement.prototype._propertyPathString):
(WebInspector.ObjectTreeArrayIndexTreeElement.prototype._updateTitle):
(WebInspector.ObjectTreeArrayIndexTreeElement.prototype._titleFragment):
(WebInspector.ObjectTreeArrayIndexTreeElement.prototype._createInteractiveGetterElement.):
(WebInspector.ObjectTreeArrayIndexTreeElement.prototype._createReadOnlyIconElement):
Index followed by formatted value. Unfortunately a page can hack up an array
with getter properties, so also support getter values in an array. This ends
up copying a lot of ObjectTreePropertyTreeElement as a result.

* UserInterface/Views/ObjectTreeMapEntryTreeElement.js: Added.
(WebInspector.ObjectTreeMapEntryTreeElement):
(WebInspector.ObjectTreeMapEntryTreeElement.prototype.get object):
(WebInspector.ObjectTreeMapEntryTreeElement.prototype._propertyPathString):
(WebInspector.ObjectTreeMapEntryTreeElement.prototype._titleFragment):
(WebInspector.ObjectTreeMapKeyTreeElement):
(WebInspector.ObjectTreeMapKeyTreeElement.prototype.displayPropertyName):
(WebInspector.ObjectTreeMapKeyTreeElement.prototype.resolvedValuePropertyPath):
(WebInspector.ObjectTreeMapValueTreeElement):
(WebInspector.ObjectTreeMapValueTreeElement.prototype.displayPropertyName):
(WebInspector.ObjectTreeMapValueTreeElement.prototype.resolvedValuePropertyPath):
Key/value followed by formatted value.

* UserInterface/Views/ObjectTreeSetIndexTreeElement.js: Added.
(WebInspector.ObjectTreeSetIndexTreeElement):
(WebInspector.ObjectTreeSetIndexTreeElement.prototype.get object):
(WebInspector.ObjectTreeSetIndexTreeElement.prototype._resolvedValuePropertyPath):
(WebInspector.ObjectTreeSetIndexTreeElement.prototype._titleFragment):
Dot followed by formatted value.

* UserInterface/Views/ObjectTreePropertyTreeElement.css:
(.object-tree-property > .titles):
Reformat.

* UserInterface/Views/ObjectTreeView.css:
(.object-tree-property :matches(.formatted-string, .formatted-regexp)):
Upgrade generic styles.

* UserInterface/Views/ObjectTreePropertyTreeElement.js:
(WebInspector.ObjectTreePropertyTreeElement.prototype._resolvedValue):
(WebInspector.ObjectTreePropertyTreeElement.prototype._resolvedValuePropertyPath):
(WebInspector.ObjectTreePropertyTreeElement.prototype._updateChildren):
(WebInspector.ObjectTreePropertyTreeElement.prototype._updateChildrenInternal):
(WebInspector.ObjectTreePropertyTreeElement.prototype._updateEntries):
(WebInspector.ObjectTreePropertyTreeElement.prototype._updateProperties):
* UserInterface/Views/ObjectTreeView.js:
(WebInspector.ObjectTreeView):
(WebInspector.ObjectTreeView.emptyMessageElement):
(WebInspector.ObjectTreeView.prototype.expand):
(WebInspector.ObjectTreeView.prototype.collapse):
(WebInspector.ObjectTreeView.prototype.update):
(WebInspector.ObjectTreeView.prototype._updateChildren):
(WebInspector.ObjectTreeView.prototype._updateEntries):
(WebInspector.ObjectTreeView.prototype._updateProperties):
(WebInspector.ObjectTreeView.prototype._handlePreviewOrTitleElementClick):
Both ObjectTreeView and ObjectTreePropertyTreeElement will fetch only collection
entries or properties depending on the type of the object being expanded.

(WebInspector.ObjectTreeView.prototype._trackWeakEntries):
(WebInspector.ObjectTreeView.prototype._untrackWeakEntries):
Allow WeakMap entries to be Garbage Collected when the ObjectTreeView
collapses or the console is cleared. FIXME for handling sub-tree WeakMaps.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180713 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoRemove unnecessary create() factory functions in CDMFoo, NamedNodeMap
gyuyoung.kim@samsung.com [Fri, 27 Feb 2015 01:04:19 +0000 (01:04 +0000)]
Remove unnecessary create() factory functions in CDMFoo, NamedNodeMap
https://bugs.webkit.org/show_bug.cgi?id=141902

Reviewed by Darin Adler.

Create instance using std::make_unique<> in CDMFactory directly. To do that,
CDMFactory uses lambdas. Additionally this patch removes NamedNodeMap::create() as well.

No new tests, no behavior changes.

* Modules/encryptedmedia/CDM.cpp:
(WebCore::installedCDMFactories):
* Modules/encryptedmedia/CDM.h:
* Modules/encryptedmedia/CDMPrivateClearKey.h:
(WebCore::CDMPrivateClearKey::CDMPrivateClearKey):
(WebCore::CDMPrivateClearKey::create): Deleted.
* Modules/encryptedmedia/CDMPrivateMediaPlayer.h:
(WebCore::CDMPrivateMediaPlayer::create): Deleted.
* dom/Element.cpp:
(WebCore::Element::attributes):
* dom/NamedNodeMap.h:
(WebCore::NamedNodeMap::NamedNodeMap):
(WebCore::NamedNodeMap::create): Deleted.
* platform/graphics/avfoundation/CDMPrivateMediaSourceAVFObjC.h:
(WebCore::CDMPrivateMediaSourceAVFObjC::create): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180712 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoThe bool returning form of BytecodeGenerator::addVar() can be removed
fpizlo@apple.com [Fri, 27 Feb 2015 00:55:19 +0000 (00:55 +0000)]
The bool returning form of BytecodeGenerator::addVar() can be removed
https://bugs.webkit.org/show_bug.cgi?id=142064

Reviewed by Mark Lam.

It's easier to implement addVar() when you don't have to return whether it's a new
variable or not.

* bytecompiler/BytecodeGenerator.cpp:
(JSC::BytecodeGenerator::addVar):
* bytecompiler/BytecodeGenerator.h:
(JSC::BytecodeGenerator::addVar): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180711 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Content Filtering] Move WebFilterEvaluator and NEFilterSource declarations to SPI...
aestes@apple.com [Fri, 27 Feb 2015 00:09:03 +0000 (00:09 +0000)]
[Content Filtering] Move WebFilterEvaluator and NEFilterSource declarations to SPI headers
https://bugs.webkit.org/show_bug.cgi?id=142062

Reviewed by Daniel Bates.

* WebCore.xcodeproj/project.pbxproj:
* platform/mac/ContentFilterMac.mm:
* platform/spi/cocoa/NEFilterSourceSPI.h: Added.
* platform/spi/cocoa/WebFilterEvaluatorSPI.h: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180708 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Win] More Debug assertion updates.
bfulgham@apple.com [Thu, 26 Feb 2015 23:53:19 +0000 (23:53 +0000)]
[Win] More Debug assertion updates.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180707 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoAdd API to remove all website data for the given data records
andersca@apple.com [Thu, 26 Feb 2015 23:34:16 +0000 (23:34 +0000)]
Add API to remove all website data for the given data records
https://bugs.webkit.org/show_bug.cgi?id=142060

Reviewed by Beth Dakin.

Source/WebCore:

* loader/cache/MemoryCache.cpp:
(WebCore::MemoryCache::removeResourcesWithOrigins):
New function that removes all resources that match the set of origins in a given session.

Source/WebKit2:

* UIProcess/API/Cocoa/_WKWebsiteDataStore.h:
Add new method declaration.

* UIProcess/API/Cocoa/_WKWebsiteDataStore.mm:
(toWebsiteDataRecords):
Add a helper function that converts an NSArray of _WKWebsiteDataRecords to a Vector of WebsiteDataRecords.

(-[_WKWebsiteDataStore removeDataOfTypes:forDataRecords:completionHandler:]):
Call through to the underlying WebsiteDataStore.

* UIProcess/WebsiteData/WebsiteDataStore.cpp:
(WebKit::WebsiteDataStore::removeData):
Figure out which processes to call deleteWebsiteDataForOrigins for and do so. Add a callback aggregator so we can
call the completion handler at the right time.

* UIProcess/WebsiteData/WebsiteDataStore.h:
Add new members.

* WebProcess/WebProcess.cpp:
(WebKit::WebProcess::deleteWebsiteDataForOrigins):
Use the newly added MemoryCache member function to delete all resources matching the set of origins.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180706 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[Win] Remove WebKitExportGenerator project (and related)
bfulgham@apple.com [Thu, 26 Feb 2015 23:26:03 +0000 (23:26 +0000)]
[Win] Remove WebKitExportGenerator project (and related)
https://bugs.webkit.org/show_bug.cgi?id=142035

Reviewed by Dean Jackson.

Source/WebKit:

* WebKit.vcxproj/WebKit.sln: Remove references to WebKitExportGenerator
* WebKit.vcxproj/WebKit.submit.sln: Ditto.
* WebKit.vcxproj/WebKit/WebKit.vcxproj: Ditto.
* WebKit.vcxproj/WebKit/WebKit.vcxproj.filters: Ditto.
* WebKit.vcxproj/WebKitExportGenerator: Removed.
* WebKit.vcxproj/WebKitExportGenerator/WebKitExportGenerator.vcxproj: Removed.
* WebKit.vcxproj/WebKitExportGenerator/WebKitExportGenerator.vcxproj.filters: Removed.
* WebKit.vcxproj/WebKitExportGenerator/WebKitExportGeneratorBuildCmd.cmd: Removed.
* WebKit.vcxproj/WebKitExportGenerator/WebKitExportGeneratorCommon.props: Removed.
* WebKit.vcxproj/WebKitExportGenerator/WebKitExportGeneratorDebug.props: Removed.
* WebKit.vcxproj/WebKitExportGenerator/WebKitExportGeneratorDebugWinCairo.props: Removed.
* WebKit.vcxproj/WebKitExportGenerator/WebKitExportGeneratorPostBuild.cmd: Removed.
* WebKit.vcxproj/WebKitExportGenerator/WebKitExportGeneratorPreBuild.cmd: Removed.
* WebKit.vcxproj/WebKitExportGenerator/WebKitExportGeneratorProduction.props: Removed.
* WebKit.vcxproj/WebKitExportGenerator/WebKitExportGeneratorRelease.props: Removed.
* WebKit.vcxproj/WebKitExportGenerator/WebKitExportGeneratorReleaseWinCairo.props: Removed.
* WebKit.vcxproj/WebKitExportGenerator/WebKitExports.def.in: Removed.
* WebKit.vcxproj/WebKitExportGenerator/make-export-file-generator: Removed.

Tools:

* win/AssembleBuildLogs/AssembleLogs.cmd: Don't look for a
build log from WebKitExportGenerator.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180705 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoRename DatabaseManager::manager() to DatabaseManager::singleton()
cdumez@apple.com [Thu, 26 Feb 2015 23:22:17 +0000 (23:22 +0000)]
Rename DatabaseManager::manager() to DatabaseManager::singleton()
https://bugs.webkit.org/show_bug.cgi?id=142054

Reviewed by Ryosuke Niwa.

Rename DatabaseManager::manager() to DatabaseManager::singleton() as
per coding style and use WTF::NeverDestroyed.

Source/WebCore:

* Modules/webdatabase/DOMWindowWebDatabase.cpp:
(WebCore::DOMWindowWebDatabase::openDatabase):
* Modules/webdatabase/DatabaseBackendBase.cpp:
(WebCore::DatabaseBackendBase::DatabaseBackendBase):
* Modules/webdatabase/DatabaseContext.cpp:
(WebCore::DatabaseContext::DatabaseContext):
(WebCore::DatabaseContext::~DatabaseContext):
(WebCore::DatabaseContext::stopDatabases):
(WebCore::DatabaseContext::databaseExceededQuota):
* Modules/webdatabase/DatabaseManager.cpp:
(WebCore::DatabaseManager::singleton):
(WebCore::DatabaseManager::manager): Deleted.
* Modules/webdatabase/DatabaseManager.h:
(WebCore::DatabaseManager::~DatabaseManager): Deleted.
* Modules/webdatabase/SQLTransactionClient.cpp:
(WebCore::SQLTransactionClient::didExceedQuota):
* history/PageCache.cpp:
(WebCore::logCanCacheFrameDecision):
(WebCore::PageCache::canCachePageContainingThisFrame):
* loader/FrameLoader.cpp:
(WebCore::FrameLoader::stopLoading):

Source/WebKit/mac:

* Storage/WebDatabaseManager.mm:
(-[WebDatabaseManager init]):
(-[WebDatabaseManager origins]):
(-[WebDatabaseManager databasesWithOrigin:]):
(-[WebDatabaseManager detailsForDatabase:withOrigin:]):
(-[WebDatabaseManager deleteAllDatabases]):
(-[WebDatabaseManager deleteOrigin:]):
(-[WebDatabaseManager deleteDatabase:withOrigin:]):
* Storage/WebDatabaseQuotaManager.mm:
(-[WebDatabaseQuotaManager usage]):
(-[WebDatabaseQuotaManager quota]):
(-[WebDatabaseQuotaManager setQuota:]):
* WebCoreSupport/WebSecurityOrigin.mm:
(-[WebSecurityOrigin usage]):
(-[WebSecurityOrigin quota]):
(-[WebSecurityOrigin setQuota:]):
* WebView/WebFrame.mm:
(-[WebFrame _cacheabilityDictionary]):
* WebView/WebView.mm:
(-[WebView _preferencesChanged:]):

Source/WebKit/win:

* WebDatabaseManager.cpp:
(WebDatabaseManager::sharedWebDatabaseManager):
(WebDatabaseManager::origins):
(WebDatabaseManager::databasesWithOrigin):
(WebDatabaseManager::detailsForDatabase):
(WebDatabaseManager::deleteAllDatabases):
(WebDatabaseManager::deleteOrigin):
(WebDatabaseManager::deleteDatabase):
(WebDatabaseManager::setQuota):
(WebKitInitializeWebDatabasesIfNecessary):
* WebSecurityOrigin.cpp:
(WebSecurityOrigin::usage):
(WebSecurityOrigin::quota):
(WebSecurityOrigin::setQuota):
* WebView.cpp:
(WebView::notifyPreferencesChanged):

Source/WebKit2:

* WebProcess/WebCoreSupport/WebChromeClient.cpp:
(WebKit::WebChromeClient::exceededDatabaseQuota):
* WebProcess/WebCoreSupport/WebDatabaseManager.cpp:
(WebKit::WebDatabaseManager::initialize):
(WebKit::WebDatabaseManager::getDatabasesByOrigin):
(WebKit::WebDatabaseManager::getDatabaseOrigins):
(WebKit::WebDatabaseManager::deleteDatabaseWithNameForOrigin):
(WebKit::WebDatabaseManager::deleteDatabasesForOrigin):
(WebKit::WebDatabaseManager::deleteAllDatabases):
(WebKit::WebDatabaseManager::setQuotaForOrigin):
* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::updatePreferences):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180704 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoVarious array access corner cases should take OSR exit feedback
fpizlo@apple.com [Thu, 26 Feb 2015 22:44:45 +0000 (22:44 +0000)]
Various array access corner cases should take OSR exit feedback
https://bugs.webkit.org/show_bug.cgi?id=142056

Reviewed by Geoffrey Garen.
Source/JavaScriptCore:

Two major changes here:

- Don't keep converting GetById into GetArrayLength if we exited due to any kind of array
  type check.

- Use a generic form of GetByVal/PutByVal if we exited due to any kind of exotic checks,
  like the Arguments safety checks. We use the "ExoticObjectMode" for out-of-bounds on
  arguments for now, since it's a convenient way of forcing out-of-bounds to be handled by
  the Generic array mode.

* bytecode/ExitKind.cpp:
(JSC::exitKindToString):
* bytecode/ExitKind.h:
* dfg/DFGArrayMode.cpp:
(JSC::DFG::ArrayMode::refine):
* dfg/DFGFixupPhase.cpp:
(JSC::DFG::FixupPhase::fixupNode):
* dfg/DFGSpeculativeJIT.cpp:
(JSC::DFG::SpeculativeJIT::compileGetByValOnArguments):
(JSC::DFG::SpeculativeJIT::compileGetArgumentsLength):
* tests/stress/array-length-array-storage-plain-object.js: Added.
(foo):
* tests/stress/array-length-plain-object.js: Added.
(foo):

LayoutTests:

* js/regress/arguments-out-of-bounds-expected.txt: Added.
* js/regress/arguments-out-of-bounds.html: Added.
* js/regress/exit-length-on-plain-object-expected.txt: Added.
* js/regress/exit-length-on-plain-object.html: Added.
* js/regress/script-tests/arguments-out-of-bounds.js: Added.
(foo):
(bar):
* js/regress/script-tests/exit-length-on-plain-object.js: Added.
(foo):
* js/regress/script-tests/string-out-of-bounds.js: Added.
(bar):
* js/regress/string-out-of-bounds-expected.txt: Added.
* js/regress/string-out-of-bounds.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180703 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agobmalloc: Large object free list can grow infinitely
ggaren@apple.com [Thu, 26 Feb 2015 22:24:37 +0000 (22:24 +0000)]
bmalloc: Large object free list can grow infinitely
https://bugs.webkit.org/show_bug.cgi?id=142055

Reviewed by Andreas Kling.

By design, we don't eagerly remove large objects from the free list.
This creates two simple pathologies:

    (1) If you free and then allocate the same object repeatedly, it will
    duplicate itself in the free list repeatedly. Since it is never
    invalid at the time of allocation, it will never be removed.

    (2) If you split and then merge the same object repeatedly, it will
    duplicate its split sibling in the free list repeatedly. If its
    sibling is in a separate free list size class, it will never be
    consulted at the time of allocation, so it will never be removed.

So, a simple "while (1) { free(malloc(x)); }" causes infinite memory
use in the free list.

The solution in this patch is a simple helper to remove garbage from the
free list if it grows too large. This pathology is not common, so the
cost is OK.

Long-term, perhaps we should rethink the laziness of these free lists.

* bmalloc/BoundaryTag.h:
(bmalloc::BoundaryTag::isMarked):
(bmalloc::BoundaryTag::setMarked): New bit, used by free list GC.

* bmalloc/FreeList.cpp:
(bmalloc::FreeList::removeInvalidAndDuplicateEntries): The GC algorithm.

* bmalloc/FreeList.h:
(bmalloc::FreeList::FreeList):
(bmalloc::FreeList::push): Invoke the GC if we're getting huge.

* bmalloc/LargeObject.h:
(bmalloc::LargeObject::isMarked):
(bmalloc::LargeObject::setMarked):
(bmalloc::LargeObject::validateSelf): Expose the new bit.

* bmalloc/Sizes.h: New constant to control GC frequency.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180701 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoFix the !ENABLE(GEOLOCATION) build after r180533
ossy@webkit.org [Thu, 26 Feb 2015 22:22:55 +0000 (22:22 +0000)]
Fix the !ENABLE(GEOLOCATION) build after r180533
https://bugs.webkit.org/show_bug.cgi?id=142053

Reviewed by Chris Dumez.

* Modules/geolocation/GeoNotifier.cpp:
* Modules/geolocation/GeoNotifier.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180700 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[EFL] Bump EFL version to 1.13.0
ossy@webkit.org [Thu, 26 Feb 2015 22:04:13 +0000 (22:04 +0000)]
[EFL] Bump EFL version to 1.13.0
https://bugs.webkit.org/show_bug.cgi?id=141708

Reviewed by Gyuyoung Kim.

* efl/jhbuild.modules:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180697 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago[iOS Media] incorrect front padding on time values
dino@apple.com [Thu, 26 Feb 2015 21:04:06 +0000 (21:04 +0000)]
[iOS Media] incorrect front padding on time values
https://bugs.webkit.org/show_bug.cgi?id=142027
<rdar://problem/19960790>

Reviewed by Brent Fulgham.

My last commit was updating the OS X file,
not the iOS file.

We don't want to prefix times with a "0".

* Modules/mediacontrols/mediaControlsApple.js:
(Controller.prototype.formatTime):
* Modules/mediacontrols/mediaControlsiOS.js:
(ControllerIOS.prototype.formatTime):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180696 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoURTBF after r180693.
ossy@webkit.org [Thu, 26 Feb 2015 20:40:57 +0000 (20:40 +0000)]
URTBF after r180693.

* CMakeLists.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180695 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years ago2015-02-26 Geoffrey Garen <ggaren@apple.com>
ggaren@apple.com [Thu, 26 Feb 2015 20:20:07 +0000 (20:20 +0000)]
2015-02-26  Geoffrey Garen  <ggaren@apple.com>

        Try to fix the Mac build.

        Unreviewed.

        * bmalloc.xcodeproj/project.pbxproj: Make FreeList.h available.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180694 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agobmalloc: Refactored SegregatedFreeList and BoundaryTag::init
ggaren@apple.com [Thu, 26 Feb 2015 20:05:14 +0000 (20:05 +0000)]
bmalloc: Refactored SegregatedFreeList and BoundaryTag::init
https://bugs.webkit.org/show_bug.cgi?id=142049

Reviewed by Anders Carlsson.

Split out a FreeList class from SegregatedFreeList. This will make it
easier to add behaviors on free list insertion and removal -- and it's
probably how I should have designed things at the start.

Moved BoundaryTag::init into LargeObject, since all the related logic
lives in LargeObject now too, and this allows us to remove BoundaryTagInlines.h.

* bmalloc.xcodeproj/project.pbxproj:
* bmalloc/BoundaryTagInlines.h: Removed.
* bmalloc/FreeList.cpp: Copied from Source/bmalloc/bmalloc/SegregatedFreeList.cpp.
(bmalloc::FreeList::takeGreedy):
(bmalloc::FreeList::take):
(bmalloc::SegregatedFreeList::SegregatedFreeList): Deleted.
(bmalloc::SegregatedFreeList::insert): Deleted.
(bmalloc::SegregatedFreeList::takeGreedy): Deleted.
(bmalloc::SegregatedFreeList::take): Deleted.
* bmalloc/FreeList.h: Copied from Source/bmalloc/bmalloc/SegregatedFreeList.h.
(bmalloc::FreeList::push):
* bmalloc/LargeObject.h:
(bmalloc::LargeObject::init):
* bmalloc/SegregatedFreeList.cpp:
(bmalloc::SegregatedFreeList::SegregatedFreeList):
(bmalloc::SegregatedFreeList::insert):
(bmalloc::SegregatedFreeList::takeGreedy):
(bmalloc::SegregatedFreeList::take):
* bmalloc/SegregatedFreeList.h:
* bmalloc/Sizes.h:
* bmalloc/VMHeap.cpp:
(bmalloc::VMHeap::grow):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180693 268f45cc-cd09-0410-ab3c-d52691b4dbfc

5 years agoDFG SSA stack accesses shouldn't speak of VariableAccessDatas
fpizlo@apple.com [Thu, 26 Feb 2015 19:51:52 +0000 (19:51 +0000)]
DFG SSA stack accesses shouldn't speak of VariableAccessDatas
https://bugs.webkit.org/show_bug.cgi?id=142036

Reviewed by Michael Saboff.

VariableAccessData is a useful thing in LoadStore and ThreadedCPS, but it's purely harmful in
SSA because you can't cook up new VariableAccessDatas. So, if you know that you want to load
or store to the stack, and you know what format to use as well as the location, then prior to
this patch you couldn't do it unless you found some existing VariableAccessData that matched
your requirements. That can be a hard task.

It's better if SSA doesn't speak of VariableAccessDatas but instead just has stack accesses
that speak of the things that a stack access needs: local, machineLocal, and format. This
patch changes the SSA way of accessing the stack to do just that.

Also add more IR validation.

* CMakeLists.txt:
* JavaScriptCore.vcxproj/JavaScriptCore.vcxproj:
* JavaScriptCore.xcodeproj/project.pbxproj:
* dfg/DFGAbstractInterpreterInlines.h:
(JSC::DFG::AbstractInterpreter<AbstractStateType>::executeEffects):
* dfg/DFGClobberize.h:
(JSC::DFG::clobberize):
* dfg/DFGConstantFoldingPhase.cpp:
(JSC::DFG::ConstantFoldingPhase::foldConstants):
* dfg/DFGDoesGC.cpp:
(JSC::DFG::doesGC):
* dfg/DFGFixupPhase.cpp:
(JSC::DFG::FixupPhase::fixupNode):
* dfg/DFGFlushFormat.h:
(JSC::DFG::isConcrete):
* dfg/DFGGraph.cpp:
(JSC::DFG::Graph::dump):
* dfg/DFGGraph.h:
* dfg/DFGMayExit.cpp:
(JSC::DFG::mayExit):
* dfg/DFGNode.cpp:
(JSC::DFG::Node::hasVariableAccessData):
* dfg/DFGNode.h:
(JSC::DFG::StackAccessData::StackAccessData):
(JSC::DFG::StackAccessData::flushedAt):
(JSC::DFG::Node::convertToPutStack):
(JSC::DFG::Node::convertToGetStack):
(JSC::DFG::Node::hasUnlinkedLocal):
(JSC::DFG::Node::hasStackAccessData):
(JSC::DFG::Node::stackAccessData):
(JSC::DFG::Node::willHaveCodeGenOrOSR):
* dfg/DFGNodeType.h:
* dfg/DFGOSRAvailabilityAnalysisPhase.cpp:
(JSC::DFG::LocalOSRAvailabilityCalculator::executeNode):
* dfg/DFGPlan.cpp:
(JSC::DFG::Plan::compileInThreadImpl):
* dfg/DFGPredictionPropagationPhase.cpp:
(JSC::DFG::PredictionPropagationPhase::propagate):
* dfg/DFGPutLocalSinkingPhase.cpp: Removed.
* dfg/DFGPutLocalSinkingPhase.h: Removed.
* dfg/DFGPutStackSinkingPhase.cpp: Copied from Source/JavaScriptCore/dfg/DFGPutLocalSinkingPhase.cpp.
(JSC::DFG::performPutStackSinking):
(JSC::DFG::performPutLocalSinking): Deleted.
* dfg/DFGPutStackSinkingPhase.h: Copied from Source/JavaScriptCore/dfg/DFGPutLocalSinkingPhase.h.
* dfg/DFGSSAConversionPhase.cpp:
(JSC::DFG::SSAConversionPhase::run):
* dfg/DFGSafeToExecute.h:
(JSC::DFG::safeToExecute):
* dfg/DFGSpeculativeJIT32_64.cpp:
(JSC::DFG::SpeculativeJIT::compile):
* dfg/DFGSpeculativeJIT64.cpp:
(JSC::DFG::SpeculativeJIT::compile):
* dfg/DFGStackLayoutPhase.cpp:
(JSC::DFG::StackLayoutPhase::run):
* dfg/DFGValidate.cpp:
(JSC::DFG::Validate::validate):
(JSC::DFG::Validate::validateCPS):
(JSC::DFG::Validate::validateSSA):
* dfg/DFGVirtualRegisterAllocationPhase.cpp:
(JSC::DFG::VirtualRegisterAllocationPhase::run):
* ftl/FTLCapabilities.cpp:
(JSC::FTL::canCompile):
* ftl/FTLLowerDFGToLLVM.cpp:
(JSC::FTL::LowerDFGToLLVM::lower):
(JSC::FTL::LowerDFGToLLVM::compileNode):
(JSC::FTL::LowerDFGToLLVM::compileGetStack):
(JSC::FTL::LowerDFGToLLVM::compilePutStack):
(JSC::FTL::LowerDFGToLLVM::compileGetLocal): Deleted.
(JSC::FTL::LowerDFGToLLVM::compilePutLocal): Deleted.
* ftl/FTLOSRExit.h:
* tests/stress/many-sunken-locals.js: Added. This failure mode was caught by some miscellaneous test, so I figured I should write an explicit test for it.
(foo):
(bar):
(baz):
(fuzz):
(buzz):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@180691 268f45cc-cd09-0410-ab3c-d52691b4dbfc