WebKit-https.git
7 years agoAssertion failure under FEImage::determineAbsolutePaintRect()
jhoneycutt@apple.com [Tue, 15 Apr 2014 05:23:07 +0000 (05:23 +0000)]
Assertion failure under FEImage::determineAbsolutePaintRect()

<https://bugs.webkit.org/show_bug.cgi?id=131660>
<rdar://problem/15669294>

Source/WebCore:
This patch merges Chromium r149536 (see
<https://chromiumcodereview.appspot.com/14701012>), which moves
m_absoluteTransform out of SVGFilter and into the base Filter class, so
that it isn't necessary to cast a Filter to SVGFilter to get the
absolute transform.

Reviewed by Geoffrey Garen.

Test: svg/filters/feImage-filter-assertion.html

* platform/graphics/filters/Filter.h:
(WebCore::Filter::Filter):
Changed to take the absolute transform.
(WebCore::Filter::absoluteTransform):
Moved from SVGFilter.
(WebCore::Filter::mapAbsolutePointToLocalPoint):
Ditto.

* rendering/FilterEffectRenderer.cpp:
(WebCore::FilterEffectRenderer::FilterEffectRenderer):
Pass a default AffineTransform() to the Filter base class.

* svg/graphics/filters/SVGFEImage.cpp:
(WebCore::FEImage::determineAbsolutePaintRect):
Use the Filter without casting it to SVGFilter.
(WebCore::FEImage::platformApplySoftware):
Ditto.

* svg/graphics/filters/SVGFilter.cpp:
(WebCore::SVGFilter::SVGFilter):
Pass the transform to the base class, and remove initialization of a
removed member var.

* svg/graphics/filters/SVGFilter.h:
Member var moved to Filter.h.

LayoutTests:
Reviewed by Geoffrey Garen.

* svg/filters/feImage-filter-assertion-expected.txt: Added.
* svg/filters/feImage-filter-assertion.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167295 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoUpdate to CodeMirror 4.0.
commit-queue@webkit.org [Tue, 15 Apr 2014 04:31:41 +0000 (04:31 +0000)]
Update to CodeMirror 4.0.
https://bugs.webkit.org/show_bug.cgi?id=130019

The CodeMirror 4.0 library adds multiple selections and the ability to
undo/redo selections. Selections are made by holding the Command (Cmd) key
and clicking into an editor, or by holding Option (Alt) and making a block
selection followed by entering a character or moving the cursor.

Patch by Jono Wells <jonowells@apple.com> on 2014-04-14
Reviewed by Timothy Hatcher.

* Scripts/update-codemirror-resources.rb:
* Scripts/update-pretty-printer.rb:
Update scripts to reflect updated locations for CodeMirror files.
No longer copies LESS mode file as this has been integrated into CSS mode
in CodeMirror 4.0.

* Tools/PrettyPrinting/codemirror.css:
* Tools/PrettyPrinting/codemirror.js:
* Tools/PrettyPrinting/css.js:
* Tools/PrettyPrinting/javascript.js:
* UserInterface/External/CodeMirror/clojure.js:
* UserInterface/External/CodeMirror/closebrackets.js:
* UserInterface/External/CodeMirror/codemirror.css:
* UserInterface/External/CodeMirror/codemirror.js:
* UserInterface/External/CodeMirror/coffeescript.js:
* UserInterface/External/CodeMirror/comment.js:
* UserInterface/External/CodeMirror/css.js:
* UserInterface/External/CodeMirror/htmlmixed.js:
* UserInterface/External/CodeMirror/javascript.js:
* UserInterface/External/CodeMirror/livescript.js:
* UserInterface/External/CodeMirror/matchbrackets.js:
* UserInterface/External/CodeMirror/overlay.js:
* UserInterface/External/CodeMirror/placeholder.js:
* UserInterface/External/CodeMirror/runmode.js:
* UserInterface/External/CodeMirror/sass.js:
* UserInterface/External/CodeMirror/searchcursor.js:
* UserInterface/External/CodeMirror/sql.js:
* UserInterface/External/CodeMirror/xml.js:
Update to CodeMirror 4.0.

* UserInterface/Main.html: Remove less.js which is now part of css.js.

* UserInterface/Views/CSSStyleDeclarationTextEditor.css:
* UserInterface/Views/CSSStyleDeclarationTextEditor.js:
(WebInspector.CSSStyleDeclarationTextEditor.prototype.):
(WebInspector.CSSStyleDeclarationTextEditor.prototype._createColorSwatches):
(WebInspector.CSSStyleDeclarationTextEditor.prototype._updateJumpToSymbolTrackingMode):
* UserInterface/Views/SyntaxHighlightingDefaultTheme.css:
Update styles to match CodeMirror changes. Update CSSStyleDeclarationTextEditor.js
to match CodeMirror API updates (doc.removeLine() has been removed, so
replaceRange() is used instead). Also the tokenTrackingController is now enabled in the
CSSStyleDeclarationTextEditor and SourceCodeTextEditor when the Option (Alt) key
is pressed instead of the Command (Cmd) key so as not to conflict with multiple
cursor placement in the CodeMirror update.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167294 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoASSERT when firing low memory warning
ggaren@apple.com [Tue, 15 Apr 2014 03:53:47 +0000 (03:53 +0000)]
ASSERT when firing low memory warning
https://bugs.webkit.org/show_bug.cgi?id=131659

Reviewed by Mark Hahnenberg.

* heap/Heap.cpp:
(JSC::Heap::deleteAllCompiledCode): Allow deleteAllCompiledCode to be
called when no GC is happening because that is what we do when a low
memory warning fires, and it is harmless.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167293 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoMallocBench should scavenge explicitly instead of waiting
ggaren@apple.com [Tue, 15 Apr 2014 03:52:39 +0000 (03:52 +0000)]
MallocBench should scavenge explicitly instead of waiting
https://bugs.webkit.org/show_bug.cgi?id=131661

Reviewed by Andreas Kling.

PerformanceTests:
* MallocBench/MallocBench.xcodeproj/project.pbxproj: Don't build mbmalloc
by default because it will overwrite any other mbmalloc you're working
with in the WebKitBuild directory.

* MallocBench/MallocBench/Benchmark.cpp:
(Benchmark::run): Scavenge explicitly instead of waiting. This is faster,
and it's the only way to get FastMalloc to scavenge. (That's a bug in
FastMalloc, but we don't want it to interfere with broader testing.)

* MallocBench/MallocBench/mbmalloc.cpp:
* MallocBench/MallocBench/mbmalloc.h: Added a scavenge implementation
for system malloc.

Source/bmalloc:
Added explicit scavenge support to bmalloc. This isn't a memory win,
since bmalloc's per-thread cache is so small. But it makes testing
simpler.

* bmalloc/Allocator.cpp:
(bmalloc::Allocator::~Allocator):
(bmalloc::Allocator::scavenge):
* bmalloc/Allocator.h:
* bmalloc/Cache.cpp:
(bmalloc::Cache::operator new):
(bmalloc::Cache::operator delete):
(bmalloc::Cache::Cache):
(bmalloc::Cache::scavenge):
* bmalloc/Cache.h:
* bmalloc/Deallocator.cpp:
(bmalloc::Deallocator::~Deallocator):
(bmalloc::Deallocator::scavenge):
* bmalloc/Deallocator.h: Factored existing scavenging code into helper
functions, for reuse.

* bmalloc/Heap.cpp:
(bmalloc::sleep):
(bmalloc::Heap::concurrentScavenge):
(bmalloc::Heap::scavenge):
(bmalloc::Heap::scavengeSmallPages):
(bmalloc::Heap::scavengeMediumPages):
(bmalloc::Heap::scavengeLargeRanges):
* bmalloc/Heap.h: Made scavenge sleep duration a parameter. Forced
scavenging -- in response to a benchmark or a low memory warning --
wants to complete as soon as possible, so its sleep duration is 0.

* bmalloc/bmalloc.h:
(bmalloc::api::scavenge):
* bmalloc/mbmalloc.cpp: Exported the scavenge API for MallocBench's use.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167292 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoREGRESSION (r158617): Find on Page can get stuck in a loop when the search string...
darin@apple.com [Tue, 15 Apr 2014 03:06:21 +0000 (03:06 +0000)]
REGRESSION (r158617): Find on Page can get stuck in a loop when the search string occurs in an <input> in a <fieldset>
https://bugs.webkit.org/show_bug.cgi?id=126322

Reviewed by Ryosuke Niwa.

One additional tweak to the fix for the bug above.
Fixes crash in editing/editability/ignored-content.html test.

* html/HTMLObjectElement.cpp:
(WebCore::HTMLObjectElement::canContainRangeEndPoint): Call through to
HTMLElement::canContainRangeEndPoint, bypassing HTMLPlugInElement override
that always returns false. Without this change, this function was always
returning false.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167291 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoCrash in TileController::tileRevalidationTimerFired
simon.fraser@apple.com [Tue, 15 Apr 2014 02:54:13 +0000 (02:54 +0000)]
Crash in TileController::tileRevalidationTimerFired
https://bugs.webkit.org/show_bug.cgi?id=131656
<rdar://problem/16583166>

Reviewed by Sam Weinig.

It's possible for the TileController revalidation timer to fire after
the GraphicsLayer has been destroyed, so the PlatformCALayer no longer
has an owningGraphicsLayer.

Bail from the timer callback if owningGraphicsLayer() is null.

Also some drive-by 0 -> nullptr changes.

* platform/graphics/ca/GraphicsLayerCA.cpp:
(WebCore::GraphicsLayerCA::willBeDestroyed):
* platform/graphics/ca/PlatformCALayer.cpp:
(WebCore::PlatformCALayer::~PlatformCALayer):
* platform/graphics/ca/mac/TileController.mm:
(WebCore::TileController::tileRevalidationTimerFired):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167290 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoUse 4kB pages on Mac
ggaren@apple.com [Tue, 15 Apr 2014 02:44:09 +0000 (02:44 +0000)]
Use 4kB pages on Mac
https://bugs.webkit.org/show_bug.cgi?id=131658

Reviewed by Sam Weinig.

This reduces memory use a lot on Membuster:

                                                          base                      patch                                Δ
        Execution Time:
            reddit_memory_warning                         18ms                       17ms                   ^ 1.06x faster
            flickr_memory_warning                         34ms                       36ms                   ! 1.06x slower
            theverge_memory_warning                       39ms                       41ms                   ! 1.05x slower

            <geometric mean>                              29ms                       29ms                   ! 1.02x slower
            <arithmetic mean>                             30ms                       31ms                   ! 1.03x slower
            <harmonic mean>                               27ms                       27ms                    ^ 1.0x faster

        Peak Memory:
            reddit_memory_warning                     16,412kB                   16,436kB                    ! 1.0x bigger
            flickr_memory_warning                     30,120kB                   30,184kB                    ! 1.0x bigger
            theverge_memory_warning                   33,408kB                   33,420kB                    ! 1.0x bigger

            <geometric mean>                          25,466kB                   25,499kB                    ! 1.0x bigger
            <arithmetic mean>                         26,647kB                   26,680kB                    ! 1.0x bigger
            <harmonic mean>                           24,181kB                   24,214kB                    ! 1.0x bigger

        Memory at End:
            reddit_memory_warning                      2,404kB                    1,920kB                  ^ 1.25x smaller
            flickr_memory_warning                      3,764kB                    3,072kB                  ^ 1.23x smaller
            theverge_memory_warning                    3,648kB                    3,132kB                  ^ 1.16x smaller

            <geometric mean>                           3,208kB                    2,644kB                  ^ 1.21x smaller
            <arithmetic mean>                          3,272kB                    2,708kB                  ^ 1.21x smaller
            <harmonic mean>                            3,139kB                    2,574kB                  ^ 1.22x smaller

* bmalloc.xcodeproj/project.pbxproj:
* bmalloc/BPlatform.h: Added.
* bmalloc/VMAllocate.h: Only use 16kB pages on iOS because the page size
is 4kB on Mac.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167289 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoemit_op_put_by_id should not emit a write barrier that filters on value
mhahnenberg@apple.com [Tue, 15 Apr 2014 02:20:17 +0000 (02:20 +0000)]
emit_op_put_by_id should not emit a write barrier that filters on value
https://bugs.webkit.org/show_bug.cgi?id=131654

Reviewed by Filip Pizlo.

The 32-bit implementation does this, and it can cause crashes if we later repatch the
code to allocate and store new Butterflies.

* jit/JITPropertyAccess.cpp:
(JSC::JIT::emitWriteBarrier): We also weren't verifying that the base was a cell on
32-bit if we were passed ShouldFilterBase. I also took the liberty of sinking the tag
load down into the if statement so that we don't do it if we're not filtering on the value.
* jit/JITPropertyAccess32_64.cpp:
(JSC::JIT::emit_op_put_by_id):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167288 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoFix the 32-bit build.
timothy_horton@apple.com [Tue, 15 Apr 2014 00:55:15 +0000 (00:55 +0000)]
Fix the 32-bit build.

* UIProcess/mac/ViewGestureControllerMac.mm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167287 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoREGRESSION (WebKit2): Programmatic scrolls in overflow-scrolling:touch don't work
simon.fraser@apple.com [Tue, 15 Apr 2014 00:47:46 +0000 (00:47 +0000)]
REGRESSION (WebKit2): Programmatic scrolls in overflow-scrolling:touch don't work
https://bugs.webkit.org/show_bug.cgi?id=131649

Reviewed by Tim Horton.

Update the UIScrollView's contentOffset when we're told that the scroll position
changed.

Also refactor slightly to have a single exception-protected block.

* UIProcess/Scrolling/ios/ScrollingTreeOverflowScrollingNodeIOS.mm:
(WebKit::ScrollingTreeOverflowScrollingNodeIOS::updateAfterChildren):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167286 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoFix the 32-bit build.
timothy_horton@apple.com [Tue, 15 Apr 2014 00:47:28 +0000 (00:47 +0000)]
Fix the 32-bit build.

* UIProcess/mac/ViewGestureControllerMac.mm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167285 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoSometimes-crash under handleSwipeGesture after closing a window or quitting
timothy_horton@apple.com [Tue, 15 Apr 2014 00:41:37 +0000 (00:41 +0000)]
Sometimes-crash under handleSwipeGesture after closing a window or quitting
https://bugs.webkit.org/show_bug.cgi?id=131648
<rdar://problem/15966106>

Reviewed by Simon Fraser.

* UIProcess/mac/ViewGestureController.h:
* UIProcess/mac/ViewGestureControllerMac.mm:
(WebKit::ViewGestureController::~ViewGestureController):
(WebKit::ViewGestureController::trackSwipeGesture):
Keep a Objective C object with a single boolean property, isCancelled, on
the ViewGestureController, and also retained by the swipe-tracking block.
When the ViewGestureController is destroyed, we set isCancelled to YES,
and the next time the block is invoked, we will cancel the swipe without
touching the destroyed ViewGestureController.

(WebKit::ViewGestureController::handleSwipeGesture):
Don't try to handle a swipe gesture if the drawing area is missing.

(WebKit::ViewGestureController::endSwipeGesture):
Clear the swipe cancellation tracker when the gesture completes.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167284 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[CSS Shapes] Remove some leftover shape-inside code
bjonesbe@adobe.com [Tue, 15 Apr 2014 00:27:13 +0000 (00:27 +0000)]
[CSS Shapes] Remove some leftover shape-inside code
https://bugs.webkit.org/show_bug.cgi?id=131641

Reviewed by Dean Jackson.

I discovered that some code had been leftover from the shape-inside
removal. This removes that leftover code.

No new tests, no behavior change.

* platform/text/BidiResolver.h:
* rendering/BidiRun.cpp:
(WebCore::BidiRun::BidiRun):
* rendering/RenderBlockFlow.h:
* rendering/RenderBlockLineLayout.cpp:
(WebCore::RenderBlockFlow::createLineBoxes):
(WebCore::RenderBlockFlow::constructLine):
(WebCore::computeExpansionForJustifiedText):
(WebCore::RenderBlockFlow::computeInlineDirectionPositionsForSegment):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167283 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[MSE][Mac] video.currentTime is sometimes negative.
jer.noble@apple.com [Tue, 15 Apr 2014 00:14:59 +0000 (00:14 +0000)]
[MSE][Mac] video.currentTime is sometimes negative.
https://bugs.webkit.org/show_bug.cgi?id=131644

Reviewed by Eric Carlson.

AVSampleBufferRenderSynchronizer will occasionally return slightly negative values
when beginning playback. Clamp the return value to 0.

* platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.mm:
(WebCore::MediaPlayerPrivateMediaSourceAVFObjC::currentTimeDouble):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167282 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[CSSExclusions] Remove FIXME referencing closed bug
bjonesbe@adobe.com [Tue, 15 Apr 2014 00:07:15 +0000 (00:07 +0000)]
[CSSExclusions] Remove FIXME referencing closed bug
https://bugs.webkit.org/show_bug.cgi?id=131645

Reviewed by Dean Jackson.

This is silly, but the comment is really confusing as it's entirely
wrong now.

No new tests, no behavior change.

* rendering/style/RenderStyle.cpp:
(WebCore::RenderStyle::changeRequiresRepaint):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167281 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoREGRESSION(r166027) Menu is shown and immediately hidden after doubletap gesture.
enrica@apple.com [Mon, 14 Apr 2014 23:47:33 +0000 (23:47 +0000)]
REGRESSION(r166027) Menu is shown and immediately hidden after doubletap gesture.
https://bugs.webkit.org/show_bug.cgi?id=131646
<rdar://problem/16614374>

Reviewed by Benjamin Poulain.

This is a problem with every gesture that changes the selection
We don't need to delay updating the selection if the selection is
being changed with a gesture.

* UIProcess/ios/WKContentViewInteraction.h:
* UIProcess/ios/WKContentViewInteraction.mm:
(-[WKContentView _didUpdateBlockSelectionWithTouch:withFlags:growThreshold:shrinkThreshold:]):
(-[WKContentView changeSelectionWithGestureAt:withGesture:withState:]):
(-[WKContentView changeSelectionWithTouchAt:withSelectionTouch:baseIsStart:]):
(-[WKContentView changeSelectionWithTouchesFrom:to:withGesture:withState:]):
(-[WKContentView changeBlockSelectionWithTouchAt:withSelectionTouch:forHandle:]):
(-[WKContentView _selectionChanged]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167280 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoBuild fixage.
akling@apple.com [Mon, 14 Apr 2014 23:22:16 +0000 (23:22 +0000)]
Build fixage.

* page/FrameView.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167279 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoAssertion failure !node || node->isElementNode() in
jhoneycutt@apple.com [Mon, 14 Apr 2014 23:08:51 +0000 (23:08 +0000)]
Assertion failure !node || node->isElementNode() in
WebCore::RenderBlock::inlineElementContinuation

https://bugs.webkit.org/show_bug.cgi?id=108829
<rdar://problem/13666405>

I can't reproduce this assertion failure, but there seems to be an
invalid assumption in RenderBlock::inlineElementContinuation() that
anything with the "isInline()" bit set is a RenderInline.

No new test because the test case in the bug does not repro for me.

Reviewed by Brent Fulgham.

* rendering/RenderBlock.cpp:
(WebCore::RenderBlock::inlineElementContinuation):
Dave Hyatt says that this function should only return RenderInline
objects (not non-RenderInline inline objects), so update the checks
from isInline() to isRenderInline() before casting with
toRenderInline().

* rendering/RenderInline.cpp:
(WebCore::RenderInline::inlineElementContinuation):
Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167278 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoDo more things under memory pressure on non-iOS platforms.
akling@apple.com [Mon, 14 Apr 2014 23:03:54 +0000 (23:03 +0000)]
Do more things under memory pressure on non-iOS platforms.
<https://webkit.org/b/131625>

Reviewed by Antti Koivisto.

Rename hasReceivedMemoryPressure() to isUnderMemoryPressure() and
make it use std::atomic<bool> instead of OSAtomic primitives.

Unmask most of the PLATFORM(IOS) blocks so all platforms can take
advantage of optimizations done while under pressure. Note that
isUnderMemoryPressure() will still always return false on platforms
other than iOS/WK1, but this will change soon.

* history/PageCache.cpp:
(WebCore::PageCache::canCache):
* loader/FrameLoader.cpp:
(WebCore::FrameLoader::commitProvisionalLoad):
* page/FrameView.cpp:
(WebCore::FrameView::willPaintContents):
(WebCore::FrameView::didPaintContents):
* platform/MemoryPressureHandler.cpp:
(WebCore::MemoryPressureHandler::MemoryPressureHandler):
* platform/MemoryPressureHandler.h:
(WebCore::MemoryPressureHandler::isUnderMemoryPressure):
* platform/cocoa/MemoryPressureHandlerCocoa.mm:
(WebCore::MemoryPressureHandler::setReceivedMemoryPressure):
(WebCore::MemoryPressureHandler::clearMemoryPressure):
(WebCore::MemoryPressureHandler::respondToMemoryPressureIfNeeded):
(WebCore::MemoryPressureHandler::hasReceivedMemoryPressure): Deleted.
* platform/graphics/FontCache.cpp:
(WebCore::FontCache::purgeInactiveFontDataIfNeeded):
* platform/ios/LegacyTileCache.mm:
(WebCore::LegacyTileCache::createTilesInActiveGrid):
* platform/ios/LegacyTileGrid.mm:
(WebCore::LegacyTileGrid::shouldUseMinimalTileCoverage):
* platform/ios/LegacyTileLayerPool.mm:
(WebCore::LegacyTileLayerPool::addLayer):
* platform/ios/TileControllerMemoryHandlerIOS.cpp:
(WebCore::TileControllerMemoryHandler::tileControllerGainedUnparentedTiles):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167277 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoLet the bundle form client know whether a change in a text field was initiated by...
mitz@apple.com [Mon, 14 Apr 2014 22:46:22 +0000 (22:46 +0000)]
Let the bundle form client know whether a change in a text field was initiated by user typing
https://bugs.webkit.org/show_bug.cgi?id=131643

Reviewed by Sam Weinig.

* WebProcess/InjectedBundle/API/Cocoa/WKWebProcessPlugInFormDelegatePrivate.h: Added
initiatedByUserTyping parameter to the delegate method.

* WebProcess/InjectedBundle/API/mac/WKWebProcessPlugInBrowserContextController.mm:
(-[WKWebProcessPlugInBrowserContextController _setFormDelegate:]): Updated overload of
textDidChangeInTextField to take the initiatedByUserTyping parameter and pass it along to
the delegate.

* WebProcess/InjectedBundle/APIInjectedBundleFormClient.h:
(API::InjectedBundle::FormClient::textDidChangeInTextField): Added initiatedByUserTyping
parameter.

* WebProcess/InjectedBundle/InjectedBundlePageFormClient.cpp:
(WebKit::InjectedBundlePageFormClient::textDidChangeInTextField): Maintained the existing
behavior of not calling the C SPI client if the change was not initiated by user typing.
* WebProcess/InjectedBundle/InjectedBundlePageFormClient.h: Updated for added parameter.

* WebProcess/WebCoreSupport/WebEditorClient.cpp:
(WebKit::WebEditorClient::textDidChangeInTextField): Changed to call the bundle client
function unconditionally, but pass along whether the change was initiated by user typing.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167276 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[Cocoa] Add a WKWebProcessPlugInFormDelegate method corresponding to willSendSubmitEvent
mitz@apple.com [Mon, 14 Apr 2014 22:44:57 +0000 (22:44 +0000)]
[Cocoa] Add a WKWebProcessPlugInFormDelegate method corresponding to willSendSubmitEvent
https://bugs.webkit.org/show_bug.cgi?id=131639

Reviewed by Tim Horton.

* WebProcess/InjectedBundle/API/Cocoa/WKWebProcessPlugInFormDelegatePrivate.h: Declared
new delegate method.
* WebProcess/InjectedBundle/API/mac/WKWebProcessPlugInBrowserContextController.mm:
(-[WKWebProcessPlugInBrowserContextController _setFormDelegate:]): Added an override
of API::InjectedBundle::FormClient::willSendSubmitEvent which calls the new delegate
method.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167275 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[iOS][WK2] Change the SPI used when starting the rotation animation
benjamin@webkit.org [Mon, 14 Apr 2014 22:39:18 +0000 (22:39 +0000)]
[iOS][WK2] Change the SPI used when starting the rotation animation
https://bugs.webkit.org/show_bug.cgi?id=131638

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-04-14
Reviewed by Tim Horton.

Having an update block where all the properties are changed is more convenient for Safari.

* UIProcess/API/Cocoa/WKWebView.mm:
(-[WKWebView _beginAnimatedResizeWithUpdates:]):
(-[WKWebView _beginAnimatedResizeToSize:obscuredInsets:minimumLayoutSizeOverride:]): Deleted.
* UIProcess/API/Cocoa/WKWebViewPrivate.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167274 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[iOS WK2] Hash table assertion closing a tab
simon.fraser@apple.com [Mon, 14 Apr 2014 22:37:07 +0000 (22:37 +0000)]
[iOS WK2] Hash table assertion closing a tab
https://bugs.webkit.org/show_bug.cgi?id=131640

Reviewed by Tim Horton.

Have RemoteLayerTreeContext keep track of all PlatformCALayerRemotes,
and clear their context pointer when it is being destroyed to avoid
calling into a deleted object later.

* WebProcess/WebPage/mac/PlatformCALayerRemote.cpp:
(WebKit::PlatformCALayerRemote::~PlatformCALayerRemote):
(WebKit::PlatformCALayerRemote::addAnimationForKey):
* WebProcess/WebPage/mac/PlatformCALayerRemote.h:
(WebKit::PlatformCALayerRemote::clearContext):
* WebProcess/WebPage/mac/RemoteLayerTreeContext.h:
* WebProcess/WebPage/mac/RemoteLayerTreeContext.mm:
(WebKit::RemoteLayerTreeContext::~RemoteLayerTreeContext):
(WebKit::RemoteLayerTreeContext::layerWasCreated):
(WebKit::RemoteLayerTreeContext::layerWillBeDestroyed):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167273 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoFunction.bind itself is too slow
oliver@apple.com [Mon, 14 Apr 2014 22:05:44 +0000 (22:05 +0000)]
Function.bind itself is too slow
https://bugs.webkit.org/show_bug.cgi?id=131636

Reviewed by Geoffrey Garen.

Source/JavaScriptCore:
Rather than forcing creation of an activation, we now store
bound function properties directly on the returned closure.
This is necessary to deal with code that creates many function
bindings, but does not call them very often.

This is a 60% speed up in the included js/regress test.

* builtins/BuiltinExecutables.cpp:
(JSC::BuiltinExecutables::createBuiltinExecutable):
* builtins/Function.prototype.js:
(bind.bindingFunction):
(bind.else.switch.case.1.bindingFunction.bindingFunction.bindingFunction.boundOversizedCallThunk):
(bind.else.switch.case.1.bindingFunction):
(bind.else.switch.case.2.bindingFunction.bindingFunction.bindingFunction.boundOversizedCallThunk):
(bind.else.switch.case.2.bindingFunction):
(bind.else.switch.case.3.bindingFunction.bindingFunction.bindingFunction.boundOversizedCallThunk):
(bind.else.switch.case.3.bindingFunction):
(bind.else.switch.bindingFunction):
(bind):
(bind.else.switch.case.1.bindingFunction.oversizedCall): Deleted.
(bind.else.switch.case.2.bindingFunction.oversizedCall): Deleted.
(bind.else.switch.case.3.bindingFunction.oversizedCall): Deleted.
* runtime/CommonIdentifiers.h:

LayoutTests:
New test, and fix bogus log in old one

* js/regress/function-bind-create-expected.html: Added.
* js/regress/function-bind-create.html: Added.
* js/regress/script-tests/function-bind-create.js: Added.
(test):
* js/regress/script-tests/function-bind.js:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167272 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoAllow dynamic changes of WKThumbnailView snapshot scale
timothy_horton@apple.com [Mon, 14 Apr 2014 21:51:17 +0000 (21:51 +0000)]
Allow dynamic changes of WKThumbnailView snapshot scale
https://bugs.webkit.org/show_bug.cgi?id=131628
<rdar://problem/16584156>

Reviewed by Beth Dakin.

* UIProcess/API/Cocoa/_WKThumbnailView.mm:
(-[_WKThumbnailView _requestSnapshotIfNeeded]):
Don't bail from requesting a snapshot just because we already had one; we want to resnapshot.
Keep track of when we bail from requesting a snapshot because we have one in-flight, so we can request it later.

(-[_WKThumbnailView _didTakeSnapshot:]):
Apply the scale to the bitmap size (whoops!).
Make CA always resize the content to fill the thumbnail view's layer, respecting aspect ratio.
Re-snapshot if we previously deferred a snapshot.

(-[_WKThumbnailView setScale:]):
If we get a scale change, request a new snapshot.

(-[_WKThumbnailView setUsesSnapshot:]):
Never apply thumbnail scale to the page if we're using snapshots. It's unnecessary,
because scale will be applied simply when painting the software snapshot.
Apply the thumbnail scale if we're going from usesSnapshot -> !usesSnapshot.

* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::snapshotAtSize):
Snapshots should use the greater of the two scales, so as to never leave a portion of the snapshot unpainted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167271 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoRenderLayerCompositor's m_layerForOverhangAreas should be offset by the
bdakin@apple.com [Mon, 14 Apr 2014 21:48:13 +0000 (21:48 +0000)]
RenderLayerCompositor's m_layerForOverhangAreas should be offset by the
topContentInset
https://bugs.webkit.org/show_bug.cgi?id=131632
-and corresponding-
<rdar://problem/16609602>

Reviewed by Tim Horton.

Offset m_layerForOverhangAreas by the topContentInset.
* rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::updateOverflowControlsLayers):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167270 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[sh4] Allow use of SubImmediates in LLINT.
julien.brianceau@gmail.com [Mon, 14 Apr 2014 21:46:18 +0000 (21:46 +0000)]
[sh4] Allow use of SubImmediates in LLINT.
https://bugs.webkit.org/show_bug.cgi?id=131608

Reviewed by Mark Lam.

Allow use of SubImmediates with const pool so the sh4 architecture can
share the arm path for setEntryAddress macro. It reduces architecture
specific code and lead to a more optimal generated code for sh4.

* llint/LowLevelInterpreter.asm:
* offlineasm/sh4.rb:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167269 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoREGRESSION (WebKit2 View Gestures): Double-tap doesn't work properly when there's...
timothy_horton@apple.com [Mon, 14 Apr 2014 21:38:00 +0000 (21:38 +0000)]
REGRESSION (WebKit2 View Gestures): Double-tap doesn't work properly when there's no element underneath the cursor
https://bugs.webkit.org/show_bug.cgi?id=131629
<rdar://problem/16192821>

Reviewed by Dan Bernstein.

* UIProcess/mac/ViewGestureControllerMac.mm:
(WebKit::ViewGestureController::didCollectGeometryForSmartMagnificationGesture):
If there's nothing under the cursor, zoom towards the cursor instead of towards 0,0.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167268 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoEliminate CachedFrame::m_mousePressNode
ap@apple.com [Mon, 14 Apr 2014 21:24:04 +0000 (21:24 +0000)]
Eliminate CachedFrame::m_mousePressNode
https://bugs.webkit.org/show_bug.cgi?id=131626

Reviewed by Brady Eidson.

I couldn't find any observable effect of this change.

* history/CachedFrame.cpp:
(WebCore::CachedFrameBase::CachedFrameBase):
(WebCore::CachedFrameBase::restore):
(WebCore::CachedFrame::clear):
* history/CachedFrame.h:
(WebCore::CachedFrame::documentLoader):
(WebCore::CachedFrame::mousePressNode): Deleted.
Eliminated m_mousePressNode, accessor, and code that reached out to EventHandler.

* page/EventHandler.cpp:
(WebCore::EventHandler::mousePressNode): Deleted.
(WebCore::EventHandler::setMousePressNode): Deleted.
* page/EventHandler.h:
(WebCore::EventHandler::setMousePressed): Deleted. This function was already unused.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167267 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoRun filter animations in the UI process with UI-side compositing
simon.fraser@apple.com [Mon, 14 Apr 2014 20:46:26 +0000 (20:46 +0000)]
Run filter animations in the UI process with UI-side compositing
https://bugs.webkit.org/show_bug.cgi?id=131199
<rdar://problem/16479487>

Source/WebCore:

Reviewed by Sam Weinig.

Add FilterOperation::clone() which is used during decoding.
Export some things.

* WebCore.exp.in:
* platform/graphics/filters/FilterOperation.h:

Source/WebKit2:

Reviewed by Sam Weinig.

* Shared/WebCoreArgumentCoders.cpp: Removed lots of WebCore::
(IPC::ArgumentCoder<SelectionRect>::decode):
(IPC::ArgumentCoder<PasteboardWebContent>::encode):
(IPC::ArgumentCoder<PasteboardWebContent>::decode):
(IPC::ArgumentCoder<PasteboardImage>::encode):
(IPC::ArgumentCoder<PasteboardImage>::decode):
(IPC::ArgumentCoder<URL>::decode):
(IPC::ArgumentCoder<UserStyleSheet>::encode):
(IPC::ArgumentCoder<UserStyleSheet>::decode):
(IPC::ArgumentCoder<UserScript>::encode):
(IPC::ArgumentCoder<UserScript>::decode):
(IPC::ArgumentCoder<ScrollableAreaParameters>::encode):
(IPC::ArgumentCoder<ScrollableAreaParameters>::decode):
(IPC::ArgumentCoder<FixedPositionViewportConstraints>::encode):
(IPC::ArgumentCoder<FixedPositionViewportConstraints>::decode):
(IPC::ArgumentCoder<StickyPositionViewportConstraints>::encode):
(IPC::ArgumentCoder<StickyPositionViewportConstraints>::decode):
(IPC::ArgumentCoder<FilterOperation>::encode):
(IPC::decodeFilterOperation):
(IPC::ArgumentCoder<FilterOperations>::encode):
(IPC::ArgumentCoder<WebCore::UserStyleSheet>::encode): Deleted.
(IPC::ArgumentCoder<WebCore::UserStyleSheet>::decode): Deleted.
(IPC::ArgumentCoder<WebCore::UserScript>::encode): Deleted.
(IPC::ArgumentCoder<WebCore::UserScript>::decode): Deleted.
(IPC::ArgumentCoder<WebCore::ScrollableAreaParameters>::encode): Deleted.
(IPC::ArgumentCoder<WebCore::ScrollableAreaParameters>::decode): Deleted.
(IPC::ArgumentCoder<WebCore::FixedPositionViewportConstraints>::encode): Deleted.
(IPC::ArgumentCoder<WebCore::FixedPositionViewportConstraints>::decode): Deleted.
(IPC::ArgumentCoder<WebCore::StickyPositionViewportConstraints>::encode): Deleted.
(IPC::ArgumentCoder<WebCore::StickyPositionViewportConstraints>::decode): Deleted.
(IPC::encodeFilterOperation): Deleted.
Add encoding/decoding support for FilterOperation.
decodeFilterOperation() has to be a bare function because of the RefPtr
out parameter.
REFERENCE filters should never be encoded, since they have CachedSVGDocumentReferences.

* Shared/WebCoreArgumentCoders.h:

* Shared/mac/RemoteLayerTreeTransaction.mm:
(WebKit::RemoteLayerTreeTextStream::operator<<):
Add logging for filter animations to the RemoteLayerTree log output.

* WebProcess/WebPage/mac/GraphicsLayerCARemote.h:
* WebProcess/WebPage/mac/GraphicsLayerCARemote.cpp:
(WebKit::GraphicsLayerCARemote::addAnimation): Deleted. We can run
filter animations now.

* WebProcess/WebPage/mac/PlatformCAAnimationRemote.h:
KeyValues have a RefPtr<WebCore::FilterOperation> now (which can't be part of
the union because we need its constructor to get called).
* WebProcess/WebPage/mac/PlatformCAAnimationRemote.mm:
(WebKit::PlatformCAAnimationRemote::KeyframeValue::encode): Encode the filter.
(WebKit::PlatformCAAnimationRemote::KeyframeValue::decode): Decode the filter.
(WebKit::PlatformCAAnimationRemote::setFromValue): Implement.
(WebKit::PlatformCAAnimationRemote::setToValue): Implement.
(WebKit::PlatformCAAnimationRemote::setValues): Implement.
(WebKit::animationValueFromKeyframeValue): Handle filters.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167266 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoFix the build.
simon.fraser@apple.com [Mon, 14 Apr 2014 20:38:26 +0000 (20:38 +0000)]
Fix the build.

* UIProcess/mac/PageClientImpl.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167265 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoUse after free in WebCore::CachedResourceHandleBase::~CachedResourceHandleBase /...
jer.noble@apple.com [Mon, 14 Apr 2014 20:37:52 +0000 (20:37 +0000)]
Use after free in WebCore::CachedResourceHandleBase::~CachedResourceHandleBase / WebCore::removeDetachedChildrenInContainer
https://bugs.webkit.org/show_bug.cgi?id=131169

Reviewed by Eric Carlson.

Invalidate the WebCoreAVFResourceLoader owned by MediaPlayerPrivateAVFoundationObjC
in its destructor, to prevent a private function being called in response to the
WebCoreAVFResourceLoader being stopped.

* platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.mm:
(WebCore::MediaPlayerPrivateAVFoundationObjC::~MediaPlayerPrivateAVFoundationObjC):
* platform/graphics/avfoundation/objc/WebCoreAVFResourceLoader.h:
* platform/graphics/avfoundation/objc/WebCoreAVFResourceLoader.mm:
(WebCore::WebCoreAVFResourceLoader::invalidate):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167264 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agomonotonicallyIncreasingTime() should only initialize its static timebaseInfo once.
mark.lam@apple.com [Mon, 14 Apr 2014 20:31:45 +0000 (20:31 +0000)]
monotonicallyIncreasingTime() should only initialize its static timebaseInfo once.
<https://webkit.org/b/131630>

Reviewed by Filip Pizlo.

The current initialization of the static field is not thread safe.

* wtf/CurrentTime.cpp:
(WTF::monotonicallyIncreasingTime):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167263 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[WK2 iOS] Scrolling to anchor links is broken
simon.fraser@apple.com [Mon, 14 Apr 2014 19:49:37 +0000 (19:49 +0000)]
[WK2 iOS] Scrolling to anchor links is broken
https://bugs.webkit.org/show_bug.cgi?id=131618
<rdar://problem/16599144>

Source/WebCore:

Reviewed by Tim Horton.

Have ScrollingTreeScrollingNode pass RequestedScrollPosition updates
to the scrolling tree, so that the scrolling tree can have custom behavior
for them if necessary.

* page/scrolling/ScrollingTree.h:
(WebCore::ScrollingTree::scrollingTreeNodeRequestsScroll):
* page/scrolling/ScrollingTreeScrollingNode.cpp:
(WebCore::ScrollingTreeScrollingNode::updateAfterChildren):
* page/scrolling/ScrollingTreeScrollingNode.h:

Source/WebKit2:

Reviewed by Tim Horton.

The RemoteScrollingTree implements scrollingTreeNodeRequestsScroll
to get informed about requested scroll position updates, and passes
them along via the RemoteScrollingCoordinatorProxy, WebPageProxy and PageClient
to the WKWebView, which performs a scroll.

* UIProcess/API/Cocoa/WKWebView.mm:
(-[WKWebView _scrollToContentOffset:WebCore::]): Scroll to content offset,
taking page scale and insets into account.
* UIProcess/API/Cocoa/WKWebViewInternal.h:
* UIProcess/CoordinatedGraphics/WebView.cpp:
(WebKit::WebView::requestScroll):
* UIProcess/CoordinatedGraphics/WebView.h:
* UIProcess/PageClient.h:
* UIProcess/Scrolling/RemoteScrollingCoordinatorProxy.cpp:
(WebKit::RemoteScrollingCoordinatorProxy::scrollingTreeNodeRequestsScroll):
Pass scrolls along to the WebPageProxy for the root node. We will also need
to handle programmatic scrolls for overflow soon.
* UIProcess/Scrolling/RemoteScrollingCoordinatorProxy.h:
* UIProcess/Scrolling/RemoteScrollingTree.cpp:
(WebKit::RemoteScrollingTree::scrollingTreeNodeRequestsScroll):
* UIProcess/Scrolling/RemoteScrollingTree.h:
* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::requestScroll):
* UIProcess/WebPageProxy.h:
* UIProcess/ios/PageClientImplIOS.h:
* UIProcess/ios/PageClientImplIOS.mm:
(WebKit::PageClientImpl::canScrollView):
(WebKit::PageClientImpl::requestScroll):
* UIProcess/mac/PageClientImpl.mm:
(WebKit::PageClientImpl::requestScroll):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167262 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoWeb Replay: memoize fallback time values for document.lastModified
burg@cs.washington.edu [Mon, 14 Apr 2014 19:45:24 +0000 (19:45 +0000)]
Web Replay: memoize fallback time values for document.lastModified
https://bugs.webkit.org/show_bug.cgi?id=131318

Reviewed by Joseph Pecoraro.

Source/WebCore:
If a document's Last-Modified header can't be found or used, then
document.lastModified is derived from the current system time or
from filesystem data, which is obviously nondeterministic.

It's better to handle this inside Document::lastModified rather than using
MemoizedDOMResult, because only the fallback case is nondeterministic.

Test: http/tests/inspector/replay/document-last-modified-fallback-value.html

* dom/Document.cpp:
(WebCore::Document::lastModified): Save or reuse memoized fallback value.
* replay/WebInputs.json: Add input DocumentLastModifiedDate.

LayoutTests:
* http/tests/inspector/replay/document-last-modified-fallback-value.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167261 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoA few MallocBench record/replay fixes
ggaren@apple.com [Mon, 14 Apr 2014 19:11:50 +0000 (19:11 +0000)]
A few MallocBench record/replay fixes
https://bugs.webkit.org/show_bug.cgi?id=131627

Reviewed by Andreas Kling.

* MallocBench/MallocBench/Interpreter.cpp:
(Interpreter::run): Accept 0-sized allocations without asserting because
WebKit does that sometimes.

* MallocBench/MallocBench/flickr.ops:
* MallocBench/MallocBench/flickr_memory_warning.ops:
* MallocBench/MallocBench/reddit.ops:
* MallocBench/MallocBench/reddit_memory_warning.ops:
* MallocBench/MallocBench/theverge.ops:
* MallocBench/MallocBench/theverge_memory_warning.ops: Updated these
recordings because a bug in the recording mechanism caused one out of
every few thousand slot values to be bogus.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167260 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[GTK] Unreviewed GTK gardening.
commit-queue@webkit.org [Mon, 14 Apr 2014 19:02:26 +0000 (19:02 +0000)]
[GTK] Unreviewed GTK gardening.

Patch by Eduardo Lima Mitev <elima@igalia.com> on 2014-04-14

* platform/gtk/TestExpectations: Update test expectations for new failing test 'editing/editability/ignored-content.html'.
* platform/gtk/inspector-protocol/dom/getAccessibilityPropertiesForNode-expected.txt: Rebaselined.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167259 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoVersioning.
bshafiei@apple.com [Mon, 14 Apr 2014 19:00:23 +0000 (19:00 +0000)]
Versioning.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167258 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoKeep secondary tile grid for zoomed-out scale
antti@apple.com [Mon, 14 Apr 2014 18:40:04 +0000 (18:40 +0000)]
Keep secondary tile grid for zoomed-out scale
https://bugs.webkit.org/show_bug.cgi?id=131586

Reviewed by Darin Adler.

Source/WebCore:
* platform/graphics/ca/GraphicsLayerCA.cpp:
(WebCore::GraphicsLayerCA::updateContentsScale):

    Don't repaint tiled backing with setNeedsDisplay, it invalidates itself correctly in setContentsScale.
    Update custom child layers when tiled backing scale changes.

* platform/graphics/ca/mac/TileController.h:
* platform/graphics/ca/mac/TileController.mm:
(WebCore::TileController::TileController):
(WebCore::TileController::setNeedsDisplay):

    Drop the whole zoomed-out grid on full repaint.

(WebCore::TileController::setNeedsDisplayInRect):

    Drop changed zoomed-out tiles. A more sophisticated strategy is possible.

(WebCore::TileController::setContentsScale):

    Swap the zoomed-out grid in and out as needed.
    Repaint the active grid after scale change so the client does not have to.

(WebCore::TileController::contentsScale):

    Get the content scale from the tile grid so it is not kept in two places.

(WebCore::TileController::zoomedOutContentsScale):
(WebCore::TileController::setZoomedOutContentsScale):

    Drop the zoomed-out grid if it no longer matches the zoomed-out scale.

(WebCore::TileController::tileRevalidationTimerFired):
(WebCore::TileController::retainedTileBackingStoreMemory):
(WebCore::TileController::containerLayers):

    Return both zoomed-out tiles and the active tiles. Active tiles are on top.

(WebCore::TileController::numberOfUnparentedTiles):
(WebCore::TileController::removeUnparentedTilesNow):
* platform/graphics/ca/mac/TileGrid.h:
* platform/graphics/ca/mac/TileGrid.mm:
(WebCore::TileGrid::dropTilesInRect):

    Add a function for dropping tiles.

(WebCore::TileGrid::revalidateTiles):

Source/WebKit2:
* WebProcess/WebPage/mac/PlatformCALayerRemoteTiledBacking.cpp:
(WebKit::PlatformCALayerRemoteTiledBacking::PlatformCALayerRemoteTiledBacking):
(WebKit::PlatformCALayerRemoteTiledBacking::customSublayers):

    Always request new sublayer list from tile controller.

* WebProcess/WebPage/mac/PlatformCALayerRemoteTiledBacking.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167256 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoArray.prototype.concat should allocate output storage only once.
akling@apple.com [Mon, 14 Apr 2014 18:04:20 +0000 (18:04 +0000)]
Array.prototype.concat should allocate output storage only once.
<https://webkit.org/b/131609>

Do a first pass across 'this' and any arguments to compute the
final size of the resulting array from Array.prototype.concat.
This avoids having to grow the output incrementally as we go.

This also includes two other micro-optimizations:

- Mark getProperty() with ALWAYS_INLINE.

- Use JSArray::length() instead of taking the generic property
  lookup path when we know an argument is an Array.

My MBP says ~3% progression on Dromaeo/jslib-traverse-jquery.

Reviewed by Oliver & Darin.

* runtime/ArrayPrototype.cpp:
(JSC::getProperty):
(JSC::arrayProtoFuncConcat):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167255 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoFixed svn:ignore on bmalloc.xcodeproj, it had erroneous leading spaces.
ap@apple.com [Mon, 14 Apr 2014 17:33:20 +0000 (17:33 +0000)]
Fixed svn:ignore on bmalloc.xcodeproj, it had erroneous leading spaces.

* bmalloc.xcodeproj: Modified property svn:ignore.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167254 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoMake WK(Web)View magnification setters actually use view-relative positions
timothy_horton@apple.com [Mon, 14 Apr 2014 17:27:59 +0000 (17:27 +0000)]
Make WK(Web)View magnification setters actually use view-relative positions
https://bugs.webkit.org/show_bug.cgi?id=131611
<rdar://problem/15965239>

Reviewed by Darin Adler.

* UIProcess/API/mac/WKView.mm:
(-[WKView setMagnification:centeredAtPoint:]):
(-[WKView setMagnification:]):
Use scalePageInViewCoordinates instead.

* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::scalePageInViewCoordinates):
* UIProcess/WebPageProxy.h:
* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::scalePageInViewCoordinates):
(WebKit::WebPage::pageScaleFactor):
* WebProcess/WebPage/WebPage.h:
* WebProcess/WebPage/WebPage.messages.in:
Add scalePageInViewCoordinates, which turns the scale centerpoint within the view
into what scalePage expects: a post-scale scroll offset.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167253 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoUnreviewed, rolling out r167249.
akling@apple.com [Mon, 14 Apr 2014 17:26:28 +0000 (17:26 +0000)]
Unreviewed, rolling out r167249.
https://bugs.webkit.org/show_bug.cgi?id=131621

broke 3 tests on cloop (Requested by kling on #webkit).

Reverted changeset:

"Array.prototype.concat should allocate output storage only
once."
https://bugs.webkit.org/show_bug.cgi?id=131609
http://trac.webkit.org/changeset/167249

Patch by Commit Queue <commit-queue@webkit.org> on 2014-04-14

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167252 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoUpdate test result
oliver@apple.com [Mon, 14 Apr 2014 17:17:20 +0000 (17:17 +0000)]
Update test result

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167251 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoFixed potential integer truncation.
commit-queue@webkit.org [Mon, 14 Apr 2014 16:31:43 +0000 (16:31 +0000)]
Fixed potential integer truncation.
https://bugs.webkit.org/show_bug.cgi?id=131615

Patch by Alex Christensen <achristensen@webkit.org> on 2014-04-14
Reviewed by Darin Adler.

* assembler/X86Assembler.h:
(JSC::X86Assembler::fillNops):
Truncate the size_t to an unsigned after it is limited to 15 instead of before.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167250 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoArray.prototype.concat should allocate output storage only once.
akling@apple.com [Mon, 14 Apr 2014 16:19:58 +0000 (16:19 +0000)]
Array.prototype.concat should allocate output storage only once.
<https://webkit.org/b/131609>

Do a first pass across 'this' and any arguments to compute the
final size of the resulting array from Array.prototype.concat.
This avoids having to grow the output incrementally as we go.

This also includes two other micro-optimizations:

- Mark getProperty() with ALWAYS_INLINE.

- Use JSArray::length() instead of taking the generic property
  lookup path when we know an argument is an Array.

My MBP says ~3% progression on Dromaeo/jslib-traverse-jquery.

Reviewed by Darin Adler.

* runtime/ArrayPrototype.cpp:
(JSC::getProperty):
(JSC::arrayProtoFuncConcat):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167249 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoOptimize Canvas fill and drawImage with SourceIn, DestinationIn, SourceOut, and Desti...
krit@webkit.org [Mon, 14 Apr 2014 16:02:03 +0000 (16:02 +0000)]
Optimize Canvas fill and drawImage with SourceIn, DestinationIn, SourceOut, and DestinationAtop using transparencyLayer.
https://bugs.webkit.org/show_bug.cgi?id=79659

Reviewed by Darin Adler.

Source/WebCore:
Optimize fill() and fillRect() operations in Canvas on composited contexts by
10 to 20 times on CG.

Replacing the ImageBuffer code by transparency layers allows the
graphics library to optimize the drawing.

Doing the same for drawImage() would give performance regressions.

An inline function will create a transparency layer for CG. Cairo graphics
does not composite correctly when a transparency layer gets created.
The inline function is just a NOOP for Cairo.

This fixes bug 131303 as well.

Added performance tests with r167124 already.

* html/canvas/CanvasRenderingContext2D.cpp:
(WebCore::CanvasRenderingContext2D::fillInternal):
(WebCore::CanvasRenderingContext2D::strokeInternal):
(WebCore::CanvasRenderingContext2D::beginCompositeLayer):
(WebCore::CanvasRenderingContext2D::endCompositeLayer):
(WebCore::CanvasRenderingContext2D::fillRect):
(WebCore::CanvasRenderingContext2D::strokeRect):
(WebCore::CanvasRenderingContext2D::drawTextInternal):
(WebCore::CanvasRenderingContext2D::fullCanvasCompositedFill): Deleted.
* html/canvas/CanvasRenderingContext2D.h:

LayoutTests:
Unskip previously failing tests.

* platform/gtk/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167248 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoLots of compositing test failures after r167152
timothy_horton@apple.com [Mon, 14 Apr 2014 15:42:28 +0000 (15:42 +0000)]
Lots of compositing test failures after r167152
https://bugs.webkit.org/show_bug.cgi?id=131574

Reviewed by Darin Adler.

* platform/graphics/GraphicsLayer.cpp:
(WebCore::dumpChildren):
(WebCore::GraphicsLayer::dumpProperties):
Make child-dumping recursive so that we can easily skip layers up to any depth.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167246 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoFix incorrect indentations in CodeGeneratorJS.pm introduced in r165521
pmolnar.u-szeged@partner.samsung.com [Mon, 14 Apr 2014 12:45:01 +0000 (12:45 +0000)]
Fix incorrect indentations in CodeGeneratorJS.pm introduced in r165521
https://bugs.webkit.org/show_bug.cgi?id=131613

Reviewed by Csaba Osztrogonác.

* bindings/scripts/CodeGeneratorJS.pm:
(GenerateImplementation):
Fixed 5-space indentation.
* bindings/scripts/test/JS/JSTestNondeterministic.cpp:
Updated the tests accordingly.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167244 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoFix commit-log-editor bug revealed by r165447
commit-queue@webkit.org [Mon, 14 Apr 2014 12:24:23 +0000 (12:24 +0000)]
Fix commit-log-editor bug revealed by r165447
https://bugs.webkit.org/show_bug.cgi?id=130676

Patch by Jozsef Berta <jberta.u-szeged@partner.samsung.com> on 2014-04-14
Reviewed by Csaba Osztrogonác.

* Scripts/commit-log-editor:
(createCommitMessage): Omitting empty Source/JavaScriptCore:... blocks.
Add \n before the first block too, because the longest common prefix now ends with only one newline.
(removeLongestCommonPrefixEndingInNewline): The longest common prefix ends with only one newline,
now the last block of the common prefix isn't duplicated below. Changing the function name accordingly.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167243 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[JSC] CSSStyleDeclaration report incorrect descriptor
rego@igalia.com [Mon, 14 Apr 2014 11:21:49 +0000 (11:21 +0000)]
[JSC] CSSStyleDeclaration report incorrect descriptor
https://bugs.webkit.org/show_bug.cgi?id=89697

Reviewed by Benjamin Poulain.

Source/WebCore:

Change descriptor of CSSStyleDeclaration properties in order to have
writable and enumerable attributes set to true. Configurable is kept to
false since the property is not deleteable.

Test: fast/dom/CSSStyleDeclaration/cssstyledeclaration-properties-descriptor.html

* bindings/js/JSCSSStyleDeclarationCustom.cpp:
(WebCore::JSCSSStyleDeclaration::getOwnPropertySlotDelegate): Only set
DontDelete attribute when creating the descriptor for
CSSStyleDeclaration properties.

LayoutTests:

Add new test to check the descriptor of CSSStyleDeclaration properties.

* fast/dom/CSSStyleDeclaration/cssstyledeclaration-properties-descriptor-expected.txt: Added.
* fast/dom/CSSStyleDeclaration/cssstyledeclaration-properties-descriptor.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167240 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[JSC] Improve the call site of string comparison in some hot path
benjamin@webkit.org [Mon, 14 Apr 2014 08:46:27 +0000 (08:46 +0000)]
[JSC] Improve the call site of string comparison in some hot path
https://bugs.webkit.org/show_bug.cgi?id=131605

Reviewed by Darin Adler.

Source/JavaScriptCore:

When resolved, the String of a JSString is never null. It can be empty but not null.
The null value is reserved for ropes but those would be resolved when getting the value.

Consequently, we should use the equal() operation that do not handle null values.
Using the StringImpl directly is already common in StringPrototype but it was not used here for some reason.

* jit/JITOperations.cpp:
* runtime/JSCJSValueInlines.h:
(JSC::JSValue::equalSlowCaseInline):
(JSC::JSValue::strictEqualSlowCaseInline):
(JSC::JSValue::pureStrictEqual):

Source/WebCore:

* dom/NodeRareData.h:
(WebCore::NodeListsNodeData::NodeListCacheMapEntryHash::equal):
We should use the right comparison operation depending on the Hash Traits.

Source/WTF:

* wtf/text/StringImpl.cpp:
(WTF::stringImplContentEqual):
Inline that function to reduce the call overhead for JSC.
This is only inlined twice, it is not catastrophic for our binary.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167220 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoMerge MemoryPressureHandler{Mac,IOS}.mm
akling@apple.com [Mon, 14 Apr 2014 08:45:23 +0000 (08:45 +0000)]
Merge MemoryPressureHandler{Mac,IOS}.mm
<https://webkit.org/b/131603>

Join the iOS and Mac platform implementations of MemoryPressureHandler
under the shared Cocoa banner. Each platform still has its own quirky
behavior, but this puts them in the same file so we can start sharing.

Reviewed by Darin Adler.

* WebCore.xcodeproj/project.pbxproj:
* platform/cocoa/MemoryPressureHandlerCocoa.mm: Renamed from Source/WebCore/platform/mac/MemoryPressureHandlerMac.mm.
(WebCore::MemoryPressureHandler::platformReleaseMemory):
(WebCore::MemoryPressureHandler::install):
(WebCore::MemoryPressureHandler::uninstall):
(WebCore::MemoryPressureHandler::holdOff):
(WebCore::MemoryPressureHandler::respondToMemoryPressure):
(WebCore::respondToMemoryPressureCallback):
(WebCore::MemoryPressureHandler::installMemoryReleaseBlock):
(WebCore::MemoryPressureHandler::setReceivedMemoryPressure):
(WebCore::MemoryPressureHandler::hasReceivedMemoryPressure):
(WebCore::MemoryPressureHandler::clearMemoryPressure):
(WebCore::MemoryPressureHandler::shouldWaitForMemoryClearMessage):
(WebCore::MemoryPressureHandler::respondToMemoryPressureIfNeeded):
* platform/ios/MemoryPressureHandlerIOS.mm: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167219 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoCSS JIT: compile the :nth-child() pseudo class
benjamin@webkit.org [Mon, 14 Apr 2014 08:42:53 +0000 (08:42 +0000)]
CSS JIT: compile the :nth-child() pseudo class
https://bugs.webkit.org/show_bug.cgi?id=131602

Reviewed by Andreas Kling.

Source/WebCore:

Tests: fast/selectors/nth-child-bounds.html
       fast/selectors/nth-child-with-backtracking.html

Compile the :nth-child() pseudo class function + some related clean up.

* css/CSSSelector.cpp:
(WebCore::CSSSelector::nthA):
(WebCore::CSSSelector::nthB):
Expose the parsed value of an+b filters. Those values are used to compile
the selector.

(WebCore::CSSSelector::RareData::parseNth):
While working on the patch, I discovered some severe issues with the parsing of large
values of a and/or b. The problem comes from the way the CSS parser handle the values:
the values are parsed as a double then converted to an AtomicString for CSSSelector.

There are many problems related to large values but we never got bug reports because
they are very uncommon. Fixing those problem would require changing the parser.

Here, CSSSelector::RareData::parseNth() is hardened a little bit to avoid absurd values
of a and b.

* css/CSSSelector.h:
* cssjit/RegisterAllocator.h:
It looks like I forgot RDX in the list of register. Add it now since it is required
for SelectorCodeGenerator::modulo().

* cssjit/SelectorCompiler.cpp:
(WebCore::SelectorCompiler::addPseudoType):
(WebCore::SelectorCompiler::SelectorCodeGenerator::SelectorCodeGenerator):
(WebCore::SelectorCompiler::SelectorCodeGenerator::modulo):
(WebCore::SelectorCompiler::SelectorCodeGenerator::moduloIsZero):
There is no modulo() operation exposed on the macro assemblers. This is a basic
implementation on top of idiv for x86_64.

Since idiv works exclusively with RAX and RDX, most of the code is about getting
those registers efficiently.

(WebCore::SelectorCompiler::SelectorCodeGenerator::generateElementMatching):
(WebCore::SelectorCompiler::setElementChildIndex):
(WebCore::SelectorCompiler::setElementChildIndexAndUpdateStyle):
(WebCore::SelectorCompiler::SelectorCodeGenerator::generateElementIsNthChild):
This is pretty much a straightforward implementation of :nth-child().
The first part counts the number of previous elements.
The second part updates the tree if this is style resolution.
The last part compares the number of previous siblings to an+b to find if the filter matches.

The only part that diverges from SelectorChecker is how childIndex is used. Instead of testing it
at every iteration, only the first iteration handle the cache.

* dom/ElementRareData.h:
(WebCore::ElementRareData::childIndexMemoryOffset):
* dom/Node.h:
(WebCore::Node::rareDataMemoryOffset):
(WebCore::Node::flagHasRareData):
* rendering/style/RenderStyle.h:

LayoutTests:

Add a couple of test for the new code:
-nth-child-with-backtracking tests the register pressure with backtracking.
-nth-child-bounds tests invalid selectors do not cause problems.

* fast/selectors/nth-child-bounds-expected.txt: Added.
* fast/selectors/nth-child-bounds.html: Added.
* fast/selectors/nth-child-with-backtracking-expected.txt: Added.
* fast/selectors/nth-child-with-backtracking.html: Added.

* http/tests/security/video-poster-cross-origin-crash.html:
Now that CSSSelector filters out ridiculously bad values, the pseudo class in this test
was no longer executed.
The particular value of nth-child is irrelevant for this test, all it needs it the tree marking
while not matching.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167218 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoSupport setting a background color on page overlays
timothy_horton@apple.com [Mon, 14 Apr 2014 08:10:22 +0000 (08:10 +0000)]
Support setting a background color on page overlays
https://bugs.webkit.org/show_bug.cgi?id=131600

Reviewed by Darin Adler.

* WebProcess/WebPage/PageOverlay.cpp:
(WebKit::PageOverlay::PageOverlay):
(WebKit::PageOverlay::bounds):
(WebKit::PageOverlay::startFadeInAnimation):
(WebKit::PageOverlay::startFadeOutAnimation):
(WebKit::PageOverlay::startFadeAnimation):
(WebKit::PageOverlay::fadeAnimationTimerFired):
Minor style adjustments.
Use more references everywhere.

(WebKit::PageOverlay::setBackgroundColor):
(WebKit::PageOverlay::setNeedsDisplay):
* WebProcess/WebPage/PageOverlay.h:
(WebKit::PageOverlay::backgroundColor):
Keep track of our background color, and push it down to the page overlay controller if we have one.

* WebProcess/WebPage/PageOverlayController.cpp:
(WebKit::updateOverlayGeometry):
(WebKit::PageOverlayController::clearPageOverlay):
Page overlays will always have the right size, and just use drawsContents to determine
whether or not they should tile/have backing store/etc.

(WebKit::PageOverlayController::installPageOverlay):
(WebKit::PageOverlayController::didChangeOverlayBackgroundColor):
Set the background color of the layer.

* WebProcess/WebPage/PageOverlayController.h:

* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::WebPage):
References!

* rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::setDocumentOverlayRootLayer):
Mark the compositing tree as needing a rebuild when we get a new document-relative
overlay layer; otherwise we were depending on something else coming along and
requiring a rebuild, which didn't always happen.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167216 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[CSS Regions] Hit testing doesn't work in video
mihnea@adobe.com [Mon, 14 Apr 2014 07:50:51 +0000 (07:50 +0000)]
[CSS Regions] Hit testing doesn't work in video
https://bugs.webkit.org/show_bug.cgi?id=131485

Reviewed by Andrei Bucur.

Source/WebCore:

When hit testing flow thread layer through the region layer,
we have to pass the depth sorting information and take that
into account for the situation in which an ancestor of the region
has preserve-3d transform style.

Test: fast/regions/hit-test-region-preserve3d-container.html

* rendering/RenderLayer.cpp:
(WebCore::RenderLayer::hitTestLayer):
(WebCore::RenderLayer::hitTestFlowThreadIfRegionForFragments):
* rendering/RenderLayer.h:

LayoutTests:

* fast/regions/hit-test-region-preserve3d-container-expected.txt: Added.
* fast/regions/hit-test-region-preserve3d-container.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167215 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoUpdate html5lib test data to latest version
commit-queue@webkit.org [Mon, 14 Apr 2014 07:19:39 +0000 (07:19 +0000)]
Update html5lib test data to latest version
https://bugs.webkit.org/show_bug.cgi?id=131588

Patch by Koji Ishii <kojishi@gmail.com> on 2014-04-14
Reviewed by Darin Adler.

This patch imports the latest version of the "dat" files from
https://github.com/html5lib/html5lib-tests
Also updated expected files for failures of new tests.

* html5lib/generated/run-adoption01-data-expected.txt: Passes now as a bug in the test was fixed.
* html5lib/generated/run-adoption01-write-expected.txt: Supressed new failures.
* html5lib/generated/run-inbody01-data-expected.txt: Added.
* html5lib/generated/run-inbody01-data.html: Added.
* html5lib/generated/run-inbody01-write-expected.txt: Added.
* html5lib/generated/run-inbody01-write.html: Added.
* html5lib/generated/run-main-element-data-expected.txt: Added.
* html5lib/generated/run-main-element-data.html: Added.
* html5lib/generated/run-main-element-write-expected.txt: Added.
* html5lib/generated/run-main-element-write.html: Added.
* html5lib/generated/run-template-data-expected.txt: Supressed new failures.
* html5lib/generated/run-template-write-expected.txt: Supressed new failures.
* html5lib/generated/run-tests21-data-expected.txt: Supressed new failures.
* html5lib/generated/run-tests21-write-expected.txt: Supressed new failures.
* html5lib/generated/run-tests25-data-expected.txt: Supressed new failures.
* html5lib/generated/run-tests25-write-expected.txt: Supressed new failures.
* html5lib/resources/adoption01.dat:
* html5lib/resources/adoption02.dat:
* html5lib/resources/comments01.dat:
* html5lib/resources/doctype01.dat:
* html5lib/resources/domjs-unsafe.dat:
* html5lib/resources/entities01.dat:
* html5lib/resources/entities02.dat:
* html5lib/resources/html5test-com.dat:
* html5lib/resources/inbody01.dat:
* html5lib/resources/isindex.dat:
* html5lib/resources/main-element.dat: Added.
* html5lib/resources/pending-spec-changes-plain-text-unsafe.dat:
* html5lib/resources/pending-spec-changes.dat:
* html5lib/resources/plain-text-unsafe.dat:
* html5lib/resources/scriptdata01.dat:
* html5lib/resources/tables01.dat:
* html5lib/resources/template.dat:
* html5lib/resources/tests1.dat:
* html5lib/resources/tests10.dat:
* html5lib/resources/tests14.dat:
* html5lib/resources/tests15.dat:
* html5lib/resources/tests16.dat:
* html5lib/resources/tests17.dat:
* html5lib/resources/tests18.dat:
* html5lib/resources/tests19.dat:
* html5lib/resources/tests2.dat:
* html5lib/resources/tests20.dat:
* html5lib/resources/tests21.dat:
* html5lib/resources/tests22.dat:
* html5lib/resources/tests23.dat:
* html5lib/resources/tests25.dat:
* html5lib/resources/tests26.dat:
* html5lib/resources/tests3.dat:
* html5lib/resources/tests4.dat:
* html5lib/resources/tests5.dat:
* html5lib/resources/tests6.dat:
* html5lib/resources/tests7.dat:
* html5lib/resources/tests8.dat:
* html5lib/resources/tests9.dat:
* html5lib/resources/tests_innerHTML_1.dat:
* html5lib/resources/tricky01.dat:
* html5lib/resources/webkit01.dat:
* html5lib/resources/webkit02.dat:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167213 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agouserVisibleString should not try to "encode" host names
darin@apple.com [Mon, 14 Apr 2014 06:58:29 +0000 (06:58 +0000)]
userVisibleString should not try to "encode" host names
https://bugs.webkit.org/show_bug.cgi?id=131587
rdar://problem/14686849

Reviewed by Alexey Proskuryakov.

Source/WebCore:

* WebCore.exp.in: Updated for WebCoreNSURLExtras argument type changes.

* platform/mac/WebCoreNSURLExtras.h: Removed unneeded code to make this
Objective-C++ header compile in plain C++ files, which we never need to do.
Added missing argument name, baseURL, and changed mysterious CFIndex arguments
to the correct type, CFURLComponentType.

* platform/mac/WebCoreNSURLExtras.mm:
(WebCore::isLookalikeCharacter): Removed the inline keyword from this, and added
more lookalike characters from the Mozilla list referenced here.
(WebCore::URLByTruncatingOneCharacterBeforeComponent): Updated argument type.
(WebCore::dataForURLComponentType): Ditto.
(WebCore::userVisibleString): Only call mapHostNames if host name decoding is
needed; no encoding here.

Tools:

* TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj: Added URLExtras.mm.

* TestWebKitAPI/Tests/mac/URLExtras.mm: Added. A few tests for WebCoreNSURLExtras.
Coverage is tiny at this point; we could add a lot more cases!

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167211 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoREGRESSION (r158617): Find on Page can get stuck in a loop when the search string...
darin@apple.com [Mon, 14 Apr 2014 06:55:15 +0000 (06:55 +0000)]
REGRESSION (r158617): Find on Page can get stuck in a loop when the search string occurs in an <input> in a <fieldset>
https://bugs.webkit.org/show_bug.cgi?id=126322

Reviewed by Ryosuke Niwa.

Source/WebCore:

* dom/Element.cpp:
(WebCore::Element::canContainRangeEndPoint): Now returns false when the role of the element
is "img". This is the same rule that's hard-coded in isRenderReplacedElement for the same
reason. Need more test coverage to make sure this role feature works consistently.

* dom/Element.h: Made canContainRangeEndPoint no longer inline since it's not just a
return statement any more.

* dom/Position.cpp:
(WebCore::Position::isCandidate): Took out code that calls isRendererReplacedElement
that was added in r158617; not needed now that we updated canContainRangeEndPoint.

* dom/Range.cpp:
(WebCore::Range::firstNode): Removed code here that called isRendererReplacedElement.
This was the wrong level to be adding editing logic, and there's a FIXME here to that
effect, which we are now deleting. This was the change that broke Find.

* editing/TextIterator.cpp: Added a comment about the redundancy between the
isRendererReplacedElement and editingIgnoresContent functions.

* html/HTMLHRElement.cpp:
(WebCore::HTMLHRElement::canContainRangeEndPoint): Call through to base class instead
of just returning true when we have child nodes. Lets Element::canContainRangeEndPoint
do its thing.
* html/HTMLHRElement.h: Ditto.

* html/HTMLObjectElement.cpp:
(WebCore::HTMLObjectElement::canContainRangeEndPoint): Call through to base class instead
of just returning true when we have fallback content. Lets Element::canContainRangeEndPoint
do its thing.
* html/HTMLObjectElement.h: Ditto.

* testing/Internals.cpp:
(WebCore::Internals::countMatchesForText): Set the limit to 1000 instead of infinite.

LayoutTests:

* editing/text-iterator/count-matches-in-form-expected.txt: Added.
* editing/text-iterator/count-matches-in-form.html: Added.

* fast/text/window-find.html: Tweaked the test a bit, making it a little easier to
see if the test hasn't even run.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167210 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoUse #include instead of #import
ljaehun.lim@samsung.com [Mon, 14 Apr 2014 05:29:47 +0000 (05:29 +0000)]
Use #include instead of #import
https://bugs.webkit.org/show_bug.cgi?id=131604

Reviewed by Darin Adler.

Fix "warning: #import is a deprecated GCC extension [-Wdeprecated]".

* TestWebKitAPI/Tests/WTF/StringView.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167209 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoUse unique_ptr for FillLayer::m_next
darin@apple.com [Mon, 14 Apr 2014 04:28:25 +0000 (04:28 +0000)]
Use unique_ptr for FillLayer::m_next
https://bugs.webkit.org/show_bug.cgi?id=75222

Reviewed by Dan Bernstein.

* css/DeprecatedStyleBuilder.cpp:
(WebCore::ApplyPropertyFillLayer::applyInheritValue):
Renamed currChild to just child and prevChild to previousChild.
Changed code to pass ownership of the new FillLayer immediately.
Changed some loops to be for loops.
(WebCore::ApplyPropertyFillLayer::applyInitialValue): Ditto.
(WebCore::ApplyPropertyFillLayer::applyValue): Ditto.

* rendering/RenderBox.cpp:
(WebCore::RenderBox::backgroundHasOpaqueTopLayer): Use reference
instead of pointer.
(WebCore::RenderBox::paintFillLayers): Ditto.
* rendering/RenderBoxModelObject.cpp:
(WebCore::RenderBoxModelObject::paintFillLayerExtended): Ditto.

* rendering/style/FillLayer.cpp:
(WebCore::FillLayer::FillLayer): Removed m_next initializer since it is now an
OwnPtr and initializes automatically. In a couple other places, changed m_next
initializer to use make_unique.
(WebCore::FillLayer::~FillLayer): Wrote loop for deletion of m_next.
(WebCore::FillLayer::operator=): Removed unneeded explicit deletion of m_next.
(WebCore::FillLayer::cullEmptyLayers): Ditto.
(WebCore::clipMax): Marked inline.
(WebCore::FillLayer::computeClipMax): Rewrote to use a loop instead of recursion.
(WebCore::FillLayer::containsImage): Ditto.
(WebCore::FillLayer::imagesAreLoaded): Ditto.
(WebCore::FillLayer::hasOpaqueImage): Rewrote to use && instead of multiple if.
(WebCore::FillLayer::hasImage): Rewrote to use a loop instead of recursion.
(WebCore::FillLayer::hasFixedImage): Ditto.

* rendering/style/FillLayer.h: Changed m_next to be a unique_ptr.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167208 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[QuickLook] Move file system-related code into WebKit
aestes@apple.com [Mon, 14 Apr 2014 02:57:33 +0000 (02:57 +0000)]
[QuickLook] Move file system-related code into WebKit
https://bugs.webkit.org/show_bug.cgi?id=131597

Reviewed by Dan Bernstein.

Source/WebCore:

QuickLookHandle should not be responsible for saving a copy of the
original document to disk as it might be running in a process that
either can't write to disk or can only write into a sandboxed
container. To account for this, we need to separate the concern of
quick look conversion from that of original document saving so that
each activity can run in the appropriate process.

Created a new interface between WebCore and WebKit by adding a client
(QuickLookHandleClient) to QuickLookHandle which is notified of incoming
bytes. A new function on FrameLoaderClient tells WebKit when a new
QuickLookHandle is created, giving WebKit the opportunity to register a
handle client. Moved the existing file system-related code as well as
code only needed by WebKit1 in QuickLookHandle into a new WebKit1
QuickLookHandleClient subclass.

* WebCore.exp.in:
* WebCore.xcodeproj/project.pbxproj: Made QuickLookHandleClient.h Private.
* loader/FrameLoaderClient.h:
(WebCore::FrameLoaderClient::didCreateQuickLookHandle): Added.
* loader/ResourceLoader.cpp:
(WebCore::ResourceLoader::didCreateQuickLookHandle): Called FrameLoaderClient::didCreateQuickLookHandle().
* loader/ResourceLoader.h:
* platform/network/ResourceHandle.h: Made m_quickLook a unique_ptr.
(WebCore::ResourceHandle::setQuickLookHandle): Changed to take a unique_ptr.
* platform/network/ResourceHandleClient.h:
(WebCore::ResourceHandleClient::didCreateQuickLookHandle): Added.
* platform/network/ios/QuickLook.h: Added m_client, gave m_converter a stronger type, and made m_nsResponse a RetainPtr.
(WebCore::QuickLookHandle::setClient): Added.
(WebCore::QuickLookHandle::firstRequestURL): Added.
(WebCore::QuickLookHandle::converter): Added.
* platform/network/ios/QuickLook.mm:
(WebCore::registerQLPreviewConverterIfNeeded):
(WebCore::createTemporaryFileForQuickLook): Made non-static.
(WebCore::emptyClient): Returned a shared empty QuickLookHandleClient.
(WebCore::QuickLookHandle::QuickLookHandle): Removed file system and WebKit1-only code.
(WebCore::QuickLookHandle::create): Changed to return a unique_ptr.
(WebCore::QuickLookHandle::nsResponse):
(WebCore::QuickLookHandle::didReceiveDataArray): Removed file system code and called QuickLookHandleClient::didReceiveDataArray() instead.
(WebCore::QuickLookHandle::didReceiveData): Removed file system code and called QuickLookHandleClient::didReceiveData() instead.
(WebCore::QuickLookHandle::didFinishLoading): Removed file system code and called QuickLookHandleClient::didFinishLoading() instead.
(WebCore::QuickLookHandle::didFail): Removed file system and WebKit1-only code, calling QuickLookHandleClient::didFail() instead.
(WebCore::QuickLookHandle::~QuickLookHandle): Removed file system and WebKit1-only code. Cleared our reference to m_client.
(WebCore::QuickLookHandle::previewFileName): Retrieved from m_converter.
(WebCore::QuickLookHandle::previewRequestURL): Ditto.
* platform/network/ios/QuickLookHandleClient.h: Added.
(WebCore::QuickLookHandleClient::~QuickLookHandleClient):
(WebCore::QuickLookHandleClient::didReceiveDataArray):
(WebCore::QuickLookHandleClient::didReceiveData):
(WebCore::QuickLookHandleClient::didFinishLoading):
(WebCore::QuickLookHandleClient::didFail):

Source/WebKit/mac:

Moved file system and WebKit1-only code from QuickLookHandle into a new
QuickLookHandleClient subclass.

* WebCoreSupport/WebFrameLoaderClient.h:
* WebCoreSupport/WebFrameLoaderClient.mm:

Source/WebKit2:

* WebProcess/Network/WebResourceLoader.h: Made m_quickLookHandle a unique_ptr.
* WebProcess/ios/WebResourceLoaderIOS.mm:
(WebKit::WebResourceLoader::setUpQuickLookHandleIfNeeded):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167207 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoRelax adoption requirements of RefCounted objects that are NeverDestroyed
aestes@apple.com [Mon, 14 Apr 2014 01:35:27 +0000 (01:35 +0000)]
Relax adoption requirements of RefCounted objects that are NeverDestroyed
https://bugs.webkit.org/show_bug.cgi?id=131593

Reviewed by Dan Bernstein.

RefCounted objects that are created by NeverDestroyed<> won't have a
RefPtr adopting them, so call relaxAdoptionRequirements().

* wtf/NeverDestroyed.h:
(WTF::NeverDestroyed::NeverDestroyed):
(WTF::NeverDestroyed::MaybeRelax::MaybeRelax):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167206 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoAdded some website recordings to MallocBench -- taken from Membuster
ggaren@apple.com [Mon, 14 Apr 2014 00:02:08 +0000 (00:02 +0000)]
Added some website recordings to MallocBench -- taken from Membuster
https://bugs.webkit.org/show_bug.cgi?id=131601

Reviewed by Ryosuke Niwa.

Added flickr, reddit, and theverge -- each recorded from Membuster's
cache, with and without sending Safari a low memory warning.

* MallocBench/MallocBench.xcodeproj/project.pbxproj:
* MallocBench/MallocBench/Benchmark.cpp:
* MallocBench/MallocBench/flickr.cpp: Added.
(benchmark_flickr):
(benchmark_flickr_memory_warning):
* MallocBench/MallocBench/flickr.h: Added.
* MallocBench/MallocBench/flickr.ops: Added.
* MallocBench/MallocBench/flickr_memory_warning.ops: Added.
* MallocBench/MallocBench/reddit.cpp: Added.
(benchmark_reddit):
(benchmark_reddit_memory_warning):
* MallocBench/MallocBench/reddit.h: Added.
* MallocBench/MallocBench/reddit.ops: Added.
* MallocBench/MallocBench/reddit_memory_warning.ops: Added.
* MallocBench/MallocBench/theverge.cpp: Added.
(benchmark_theverge):
(benchmark_theverge_memory_warning):
* MallocBench/MallocBench/theverge.h: Added.
* MallocBench/MallocBench/theverge.ops: Added.
* MallocBench/MallocBench/theverge_memory_warning.ops: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167205 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoFixed some mbmalloc exports
ggaren@apple.com [Sun, 13 Apr 2014 23:28:59 +0000 (23:28 +0000)]
Fixed some mbmalloc exports
https://bugs.webkit.org/show_bug.cgi?id=131599

Reviewed by Ryosuke Niwa.

* bmalloc.xcodeproj/project.pbxproj: Made some headers a private part
of the project, so we can call them from API.

* bmalloc/mbmalloc.cpp: Marked the mbmalloc functions with default
visibility, so they show up as exported in the .dylib.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167204 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoMallocBench record/replay should support realloc
ggaren@apple.com [Sun, 13 Apr 2014 23:21:20 +0000 (23:21 +0000)]
MallocBench record/replay should support realloc
https://bugs.webkit.org/show_bug.cgi?id=131598

Reviewed by Ryosuke Niwa.

* MallocBench/MallocBench.xcodeproj/project.pbxproj: Fixed some linkage
issues that caused us not to fully link to system malloc in the default
case. Also marked mbmalloc.dylib as required so the error message will
be clearer if we mess up.

* MallocBench/MallocBench/Interpreter.cpp:
(Interpreter::run):
* MallocBench/MallocBench/Interpreter.h: Added the realloc case, and
upgraded one-letter names to full words.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167203 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoREGRESSION (r167164): -[WKNSURLSessionLocal _getCookieHeadersForTask:completionHandle...
aestes@apple.com [Sun, 13 Apr 2014 20:02:13 +0000 (20:02 +0000)]
REGRESSION (r167164): -[WKNSURLSessionLocal _getCookieHeadersForTask:completionHandler:] _block_invoke can access a deallocated completionHandler
https://bugs.webkit.org/show_bug.cgi?id=131595

Reviewed by Dan Bernstein.

Make a copy of completionHandler and release it after we're done with it.

* Shared/mac/CookieStorageShim.mm:
(-[WKNSURLSessionLocal _getCookieHeadersForTask:completionHandler:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167202 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoDon't use ImportanceAssertion on iOS
psolanki@apple.com [Sun, 13 Apr 2014 19:23:43 +0000 (19:23 +0000)]
Don't use ImportanceAssertion on iOS
https://bugs.webkit.org/show_bug.cgi?id=131481
<rdar://problem/16575830>

Reviewed by Darin Adler.

We have other API to mark processes as being in use on iOS. No need to use ImportanceAssertion.

* Platform/IPC/MessageDecoder.cpp:
* Platform/IPC/MessageDecoder.h:
* Platform/IPC/mac/ConnectionMac.cpp:
(IPC::Connection::receiveSourceEventHandler):
* Platform/IPC/mac/ImportanceAssertion.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167201 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoMove early return out of dispatch_async() block so we can return from willSendRequest...
psolanki@apple.com [Sun, 13 Apr 2014 19:20:11 +0000 (19:20 +0000)]
Move early return out of dispatch_async() block so we can return from willSendRequest quickly
https://bugs.webkit.org/show_bug.cgi?id=131478
<rdar://problem/16575535>

Reviewed by Alexey Proskuryakov.

Do a quick check to see if we need to synthesize the redirect response on the dispatch queue
and return from willSendRequest callback quickly instead of always doing an effectively synchronous
call to the main thread. We can't call synthesizeRedirectResponseIfNecessary on the dispatch
queue since that accesses the ResourceRequest.

No new tests because no change in functionality.

* platform/network/cf/ResourceHandleCFURLConnectionDelegate.h:
* platform/network/cf/ResourceHandleCFURLConnectionDelegateWithOperationQueue.cpp:
(WebCore::ResourceHandleCFURLConnectionDelegateWithOperationQueue::setupRequest): Save the
request scheme to use later for early return from willSendRequest.
(WebCore::ResourceHandleCFURLConnectionDelegateWithOperationQueue::willSendRequest):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167200 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoRewrite Function.bind as a builtin
oliver@apple.com [Sun, 13 Apr 2014 18:01:54 +0000 (18:01 +0000)]
Rewrite Function.bind as a builtin
https://bugs.webkit.org/show_bug.cgi?id=131083

Reviewed by Geoffrey Garen.

Source/JavaScriptCore:

This change removes the existing function.bind implementation
entirely so JSBoundFunction is no more.

Instead we just return a regular JS closure with a few
private properties hanging off it that allow us to perform
the necessary bound function fakery.  While most of this is
simple, a couple of key changes:

- The parser and lexer now directly track whether they're
  parsing code for call or construct and convert the private
  name @IsConstructor into TRUETOK or FALSETOK as appropriate.
  This automatically gives us the ability to vary behaviour
  from within the builtin. It also leaves a lot of headroom
  for trivial future improvements.
- The instanceof operator now uses the prototypeForHasInstance
  private name, and we have a helper function to ensure that
  all objects that need to can update their magical 'prototype'
  property pair correctly.

* API/JSScriptRef.cpp:
(parseScript):
* JavaScriptCore.xcodeproj/project.pbxproj:
* builtins/BuiltinExecutables.cpp:
(JSC::BuiltinExecutables::createBuiltinExecutable):
* builtins/Function.prototype.js:
(bind.bindingFunction):
(bind.else.bindingFunction):
(bind):
* bytecode/UnlinkedCodeBlock.cpp:
(JSC::generateFunctionCodeBlock):
* bytecompiler/NodesCodegen.cpp:
(JSC::InstanceOfNode::emitBytecode):
* interpreter/Interpreter.cpp:
* parser/Lexer.cpp:
(JSC::Lexer<T>::Lexer):
(JSC::Lexer<LChar>::parseIdentifier):
(JSC::Lexer<UChar>::parseIdentifier):
* parser/Lexer.h:
* parser/Parser.cpp:
(JSC::Parser<LexerType>::Parser):
(JSC::Parser<LexerType>::parseInner):
* parser/Parser.h:
(JSC::parse):
* parser/ParserModes.h:
* runtime/CodeCache.cpp:
(JSC::CodeCache::getGlobalCodeBlock):
(JSC::CodeCache::getFunctionExecutableFromGlobalCode):
* runtime/CommonIdentifiers.h:
* runtime/Completion.cpp:
(JSC::checkSyntax):
* runtime/Executable.cpp:
(JSC::ProgramExecutable::checkSyntax):
* runtime/FunctionPrototype.cpp:
(JSC::FunctionPrototype::addFunctionProperties):
(JSC::functionProtoFuncBind): Deleted.
* runtime/JSBoundFunction.cpp: Removed.
* runtime/JSBoundFunction.h: Removed.
* runtime/JSFunction.cpp:
(JSC::RetrieveCallerFunctionFunctor::RetrieveCallerFunctionFunctor):
(JSC::RetrieveCallerFunctionFunctor::operator()):
(JSC::retrieveCallerFunction):
(JSC::JSFunction::getOwnPropertySlot):
(JSC::JSFunction::defineOwnProperty):
* runtime/JSGlobalObject.cpp:
(JSC::JSGlobalObject::reset):
* runtime/JSGlobalObjectFunctions.cpp:
(JSC::globalFuncSetTypeErrorAccessor):
* runtime/JSGlobalObjectFunctions.h:
* runtime/JSObject.h:
(JSC::JSObject::inlineGetOwnPropertySlot):

Source/WebCore:

Switch WebCore to use the helper functions when defining the
prototype properties on DOM constructors, and update bindings
tests accordingly.

* bindings/js/JSImageConstructor.cpp:
(WebCore::JSImageConstructor::finishCreation):
* bindings/scripts/CodeGeneratorJS.pm:
(GenerateConstructorHelperMethods):
* bindings/scripts/test/JS/JSTestActiveDOMObject.cpp:
(WebCore::JSTestActiveDOMObjectConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestCustomNamedGetter.cpp:
(WebCore::JSTestCustomNamedGetterConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestEventConstructor.cpp:
(WebCore::JSTestEventConstructorConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestEventTarget.cpp:
(WebCore::JSTestEventTargetConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestException.cpp:
(WebCore::JSTestExceptionConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestGenerateIsReachable.cpp:
(WebCore::JSTestGenerateIsReachableConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestInterface.cpp:
(WebCore::JSTestInterfaceConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestMediaQueryListListener.cpp:
(WebCore::JSTestMediaQueryListListenerConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestNamedConstructor.cpp:
(WebCore::JSTestNamedConstructorConstructor::finishCreation):
(WebCore::JSTestNamedConstructorNamedConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestNode.cpp:
(WebCore::JSTestNodeConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestObj.cpp:
(WebCore::JSTestObjConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestOverloadedConstructors.cpp:
(WebCore::JSTestOverloadedConstructorsConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestSerializedScriptValueInterface.cpp:
(WebCore::JSTestSerializedScriptValueInterfaceConstructor::finishCreation):
* bindings/scripts/test/JS/JSTestTypedefs.cpp:
(WebCore::JSTestTypedefsConstructor::finishCreation):
* bindings/scripts/test/JS/JSattribute.cpp:
(WebCore::JSattributeConstructor::finishCreation):
* bindings/scripts/test/JS/JSreadonly.cpp:
(WebCore::JSreadonlyConstructor::finishCreation):

LayoutTests:

Testing.

* js/dom/function-bind-expected.txt:
* js/regress/function-bind-expected.txt: Added.
* js/regress/function-bind.html: Added.
* js/regress/script-tests/function-bind.js: Added.
(foo):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167199 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[iOS WK2] Hook up scrolling tree nodes when coming out of the page cache
simon.fraser@apple.com [Sun, 13 Apr 2014 16:41:29 +0000 (16:41 +0000)]
[iOS WK2] Hook up scrolling tree nodes when coming out of the page cache
https://bugs.webkit.org/show_bug.cgi?id=131577

Reviewed by Tim Horton.

The call to scrollingCoordinator->frameViewRootLayerDidChange() was inside
a #if !PLATFORM(IOS) block, but now that we use the ScrollingCoordinator
for WK2 we want to call this.

* loader/HistoryController.cpp:
(WebCore::HistoryController::restoreScrollPositionAndViewState):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167198 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoUnreviewed build fix after r167196.
zandobersek@gmail.com [Sun, 13 Apr 2014 14:59:41 +0000 (14:59 +0000)]
Unreviewed build fix after r167196.

* platform/RemoteCommandListener.cpp:
(WebCore::RemoteCommandListener::create): Fall back to using the new operator
for allocating RemoteCommandListener object. Using std::make_unique() requires
for the operator to be public, which doesn't work well with the static create()
method.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167197 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoRemove unnecessary uses of std::move() in return statements
zandobersek@gmail.com [Sun, 13 Apr 2014 14:18:48 +0000 (14:18 +0000)]
Remove unnecessary uses of std::move() in return statements
https://bugs.webkit.org/show_bug.cgi?id=131457

Reviewed by Darin Adler.

Don't use std::move() in return statements unless necessary as it inhibits
named return value optimizations as performed by compilers.

* Modules/battery/BatteryManager.cpp:
(WebCore::BatteryManager::create):
* html/FormController.cpp:
(WebCore::FormController::createSavedFormStateMap):
* html/canvas/WebGLRenderingContext.cpp:
(WebCore::WebGLRenderingContext::create):
* platform/RemoteCommandListener.cpp:
(WebCore::RemoteCommandListener::create):
* platform/graphics/ca/GraphicsLayerCA.cpp:
(WebCore::GraphicsLayer::create):
* platform/ios/RemoteCommandListenerIOS.mm:
(WebCore::RemoteCommandListener::create):
* rendering/RenderGrid.cpp:
(WebCore::RenderGrid::GridIterator::nextEmptyGridArea):
* rendering/RenderLayer.cpp:
(WebCore::RenderLayer::setupFilters):
* rendering/style/CounterDirectives.cpp:
(WebCore::clone):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167196 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoUnreviewed, rolling out r167168 and r167194.
commit-queue@webkit.org [Sun, 13 Apr 2014 11:23:23 +0000 (11:23 +0000)]
Unreviewed, rolling out r167168 and r167194.
https://bugs.webkit.org/show_bug.cgi?id=131589

Caused massive ASSERTION failures on the GTK Debug bot
(Requested by philn on #webkit).

Reverted changesets:

"[GTK] Add HighDPI support for non-accelerated compositing
contents"
https://bugs.webkit.org/show_bug.cgi?id=131562
http://trac.webkit.org/changeset/167168

Source/WebCore:

"Unreviewed. Fix GTK+ build with recent cairo and GTK+ after
r167168."
http://trac.webkit.org/changeset/167194

Source/WebKit2:

"Unreviewed. Fix GTK+ build with recent cairo and GTK+ after
r167168."
http://trac.webkit.org/changeset/167194

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167195 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoUnreviewed. Fix GTK+ build with recent cairo and GTK+ after r167168.
carlosgc@webkit.org [Sun, 13 Apr 2014 09:46:34 +0000 (09:46 +0000)]
Unreviewed. Fix GTK+ build with recent cairo and GTK+ after r167168.

* UIProcess/API/gtk/WebKitWebViewBase.cpp:
(webkitWebViewBaseCreateWebPage):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167194 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[GStreamer] No CORS support for media elements
commit-queue@webkit.org [Sun, 13 Apr 2014 09:33:30 +0000 (09:33 +0000)]
[GStreamer] No CORS support for media elements
https://bugs.webkit.org/show_bug.cgi?id=99037

Patch by Youenn Fablet <youenn.fablet@crf.canon.fr> on 2014-04-13
Reviewed by Philippe Normand.

Source/WebCore:

Added CORS access control check to media sources when crossorigin attribute is set.

Added getter to CORS access control check status (used to compute whether the stream is tainted or not).
Related test is http/tests/security/video-cross-origin-readback.html.

Disabled access to cross-origin streams that fail CORS check when crossorigin attribute is set.
Related test is http/tests/security/video-cross-origin-accessfailure.html.

Tests: http/tests/security/video-cross-origin-accessfailure.html
       http/tests/security/video-cross-origin-accesssameorigin.html

* platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.cpp:
(WebCore::MediaPlayerPrivateGStreamer::didPassCORSAccessCheck): Return whether media is cross-origin (tainted) or not by querying the gstreamer source layer.
* platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.h: Added MediaPlayerPrivateGStreamer::didPassCORSAccessCheck declaration.
* platform/graphics/gstreamer/WebKitWebSourceGStreamer.cpp:
(webKitWebSrcStart): Passed CORS mode parameter to the streaming client. In case of CORS check failure, stop the resource loading.
(webKitSrcPassedCORSAccessCheck): Return whether CORS access control check was done and successful.
(StreamingClient::handleResponseReceived): Take a parameter to assign the CORS access control check result.
(CachedResourceStreamingClient::CachedResourceStreamingClient): Updated setting of the ResourceLoaderOptions according CORS mode.
(CachedResourceStreamingClient::responseReceived): Check CORS and pass result to handleResponseReceived.
(ResourceHandleStreamingClient::didReceiveResponse): No CORS check.
* platform/graphics/gstreamer/WebKitWebSourceGStreamer.h: Added webKitSrcPassedCORSAccessCheck declaration.

LayoutTests:

http/tests/security/video-cross-origin-accessfailure.html verifies that cross-origin streams that fail CORS check
are not played when crossorigin attribute is set.
 http/tests/security/video-cross-origin-accesssameorigin.html verifies that access to same-origin streams
are played when crossorigin attribute is set.

* http/tests/security/video-cross-origin-accessfailure-expected.txt: Added.
* http/tests/security/video-cross-origin-accessfailure.html: Added.
* http/tests/security/video-cross-origin-accesssameorigin-expected.txt: Added.
* http/tests/security/video-cross-origin-accesssameorigin.html: Added.
* platform/efl/TestExpectations: Enabled http/tests/security/video-cross-origin-readback.html.
* platform/gtk/TestExpectations: Ditto.
* platform/mac/TestExpectations: Disabled http/tests/security/video-cross-origin-accessfailure.html.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167193 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoREGRESSION (r166860): ASSERTION FAILED: !isCalculated() on fast/css/image-set-value...
darin@apple.com [Sun, 13 Apr 2014 08:05:57 +0000 (08:05 +0000)]
REGRESSION (r166860): ASSERTION FAILED: !isCalculated() on fast/css/image-set-value-not-removed-crash.html
https://bugs.webkit.org/show_bug.cgi?id=131480

Reviewed by Andreas Kling.

Source/WebCore:

Fixes intermittent assertion failure in fast/css/image-set-value-not-removed-crash.html.

* css/CSSComputedStyleDeclaration.cpp:
(WebCore::valueForImageSliceSide): Added. Helper used below in valueForNinePieceImageSlice.
Handles calculated values by returning 0; incorrect but predictable.
(WebCore::valueForNinePieceImageSlice): Updated to call valueForImageSliceSide.
(WebCore::positionOffsetValue): Use nullptr.
(WebCore::ComputedStyleExtractor::propertyValue): Updated to call positionOffsetValue
by its new name. Removed "get" from the name.
(WebCore::positionOffsetValue): Renamed from getPositionOffsetValue.

* platform/Length.h: Made isCalculated public.

LayoutTests:

* platform/mac/TestExpectations: Unskip the test now that the assertion is fixed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167192 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoMake slow-stress tests run faster by running fewer VM variants.
fpizlo@apple.com [Sun, 13 Apr 2014 01:50:18 +0000 (01:50 +0000)]
Make slow-stress tests run faster by running fewer VM variants.

Rubber stamped by Geoffrey Garen.

Tools:

* Scripts/run-javascriptcore-tests:
* Scripts/run-jsc-stress-tests:

LayoutTests:

* jsc-layout-tests.yaml:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167191 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoFix the iOS build after r167183.
aestes@apple.com [Sat, 12 Apr 2014 23:59:35 +0000 (23:59 +0000)]
Fix the iOS build after r167183.

* platform/network/ResourceHandle.h:
* platform/network/cf/ResourceHandleCFNet.cpp:
(WebCore::ResourceHandle::schedule):
(WebCore::ResourceHandle::unschedule):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167190 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoMath.fround() should be an intrinsic
fpizlo@apple.com [Sat, 12 Apr 2014 23:01:33 +0000 (23:01 +0000)]
Math.fround() should be an intrinsic
https://bugs.webkit.org/show_bug.cgi?id=131583

Source/JavaScriptCore:

Reviewed by Geoffrey Garen.

Makes programs that use Math.fround() run up to 6x faster.

* dfg/DFGAbstractInterpreterInlines.h:
(JSC::DFG::AbstractInterpreter<AbstractStateType>::executeEffects):
* dfg/DFGByteCodeParser.cpp:
(JSC::DFG::ByteCodeParser::handleIntrinsic):
* dfg/DFGCSEPhase.cpp:
(JSC::DFG::CSEPhase::performNodeCSE):
* dfg/DFGClobberize.h:
(JSC::DFG::clobberize):
* dfg/DFGFixupPhase.cpp:
(JSC::DFG::FixupPhase::fixupNode):
* dfg/DFGNodeType.h:
* dfg/DFGPredictionPropagationPhase.cpp:
(JSC::DFG::PredictionPropagationPhase::propagate):
* dfg/DFGSafeToExecute.h:
(JSC::DFG::safeToExecute):
* dfg/DFGSpeculativeJIT32_64.cpp:
(JSC::DFG::SpeculativeJIT::compile):
* dfg/DFGSpeculativeJIT64.cpp:
(JSC::DFG::SpeculativeJIT::compile):
* ftl/FTLCapabilities.cpp:
(JSC::FTL::canCompile):
* ftl/FTLLowerDFGToLLVM.cpp:
(JSC::FTL::LowerDFGToLLVM::compileNode):
(JSC::FTL::LowerDFGToLLVM::compileArithFRound):
* runtime/Intrinsic.h:
* runtime/MathObject.cpp:
(JSC::MathObject::finishCreation):

LayoutTests:

Reviewed by Geoffrey Garen.

This test runs 4.4635x faster with the intrinsic.

* js/regress/fround-expected.txt: Added.
* js/regress/fround.html: Added.
* js/regress/script-tests/fround.js: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167189 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[iOS] Move QuickLookHandle from ResourceLoader to WebResourceLoader
aestes@apple.com [Sat, 12 Apr 2014 21:06:17 +0000 (21:06 +0000)]
[iOS] Move QuickLookHandle from ResourceLoader to WebResourceLoader
https://bugs.webkit.org/show_bug.cgi?id=131580

Reviewed by Darin Adler.

There's no need to bloat WebKit1's ResourceLoader with a pointer that
only WebKit2's WebResourceLoader cares about.

Source/WebCore:

* loader/ResourceLoader.h:
(WebCore::ResourceLoader::quickLookHandle): Deleted.
(WebCore::ResourceLoader::setQuickLookHandle): Deleted.

Source/WebKit2:

* WebProcess/Network/WebResourceLoader.cpp:
(WebKit::WebResourceLoader::didReceiveResponseWithCertificateInfo):
(WebKit::WebResourceLoader::didReceiveData):
(WebKit::WebResourceLoader::didFinishResourceLoad):
(WebKit::WebResourceLoader::didFailResourceLoad):
* WebProcess/Network/WebResourceLoader.h:
* WebProcess/ios/WebResourceLoaderIOS.mm:
(WebKit::WebResourceLoader::setUpQuickLookHandleIfNeeded):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167188 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoFTL should use stackmap register liveness
fpizlo@apple.com [Sat, 12 Apr 2014 20:41:41 +0000 (20:41 +0000)]
FTL should use stackmap register liveness
https://bugs.webkit.org/show_bug.cgi?id=130791

Reviewed by Goeffrey Garen.

Enable the stackmap register liveness support by fixing the two last bugs:

- If everything is dead after the patchpoint - a good possibility for a put_by_id -
  then we shouldn't crash due to a null scratch buffer.

- Always consider callee-saves as if they were live. More precisely, we should
  consider those callee-saves that are not saved by the enclosing function to be live.
  For now we do the much simpler thing and consider callee-saves to be always live
  since it has minimal impact on the scratch register allocator. It will know not to
  preserve those for calls, anyway.

I tried writing a test for the null scratch buffer thing, but failed. I will land the
test anyway since it seems useful.

* ftl/FTLCompile.cpp:
(JSC::FTL::usedRegistersFor):
* jit/ScratchRegisterAllocator.cpp:
(JSC::ScratchRegisterAllocator::preserveUsedRegistersToScratchBufferForCall):
(JSC::ScratchRegisterAllocator::restoreUsedRegistersFromScratchBufferForCall):
* runtime/Options.h:
* tests/stress/repeated-put-by-id-reallocating-transition.js: Added.
(foo):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167187 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoAX: Cleanup AccessibilityObject::getAttribute
cfleizach@apple.com [Sat, 12 Apr 2014 20:06:14 +0000 (20:06 +0000)]
AX: Cleanup AccessibilityObject::getAttribute
https://bugs.webkit.org/show_bug.cgi?id=131555

Reviewed by Darin Adler.

No new functionality.

* accessibility/AccessibilityObject.cpp:
(WebCore::AccessibilityObject::getAttribute):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167186 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[SOUP] Libsoup internal credential setting should be controlled by loader decision
commit-queue@webkit.org [Sat, 12 Apr 2014 20:01:45 +0000 (20:01 +0000)]
[SOUP] Libsoup internal credential setting should be controlled by loader decision
https://bugs.webkit.org/show_bug.cgi?id=130963

Patch by Youenn Fablet <youenn.fablet@crf.canon.fr> on 2014-04-12
Reviewed by Darin Adler.

Source/WebCore:

Disabled libsoup internal authentication manager for messages for which no credential is available and no stored credentials should be used.
Updated synchronous loader to return whether using credentials or not according StoredCredential loader option parameter.
Unskipped test http/tests/xmlhttprequest/cross-origin-no-authorization.html covers the patch.

* platform/network/ResourceHandleInternal.h:
(WebCore::ResourceHandleInternal::ResourceHandleInternal): Added m_useAuthenticationManager boolean to control whether disable authentication manager or not.
* platform/network/soup/ResourceHandleSoup.cpp:
(WebCore::WebCoreSynchronousLoader::WebCoreSynchronousLoader): Added m_storedCredentials member.
(WebCore::WebCoreSynchronousLoader::shouldUseCredentialStorage): Return true if stored credentials are allowed.
(WebCore::applyAuthenticationToRequest): Set m_useAuthenticationManager value to disable authentication manager if cannot use stored credentials and ResourceHandleInternal has no username and password.
(WebCore::createSoupMessageForHandleAndRequest): Disable authentication mananger according m_useAuthenticationManager value.
(WebCore::ResourceHandle::platformLoadResourceSynchronously): Added StoredCredentials loader option to the sync loader constructor.

Source/WebKit/efl:

* WebCoreSupport/FrameLoaderClientEfl.cpp:
(WebCore::FrameLoaderClientEfl::shouldUseCredentialStorage): Similarly to GTK, let soup/loader layer handle when to use credential storage. Return always true

LayoutTests:

* platform/efl/TestExpectations: Unskipped http/tests/xmlhttprequest/cross-origin-no-authorization.html.
* platform/gtk/TestExpectations: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167185 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago(before-mar-1-1000) In correct date creation prior to 1 Mar 1000
commit-queue@webkit.org [Sat, 12 Apr 2014 19:34:04 +0000 (19:34 +0000)]
(before-mar-1-1000) In correct date creation prior to 1 Mar 1000
https://bugs.webkit.org/show_bug.cgi?id=129308

Patch by Tibor Meszaros <tmeszaros.u-szeged@partner.samsung.com> on 2014-04-12
Reviewed by Darin Adler.

* js/date-constructor-expected.txt:
* js/script-tests/date-constructor.js:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167184 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoSome small loader refinements and refactoring
darin@apple.com [Sat, 12 Apr 2014 19:10:56 +0000 (19:10 +0000)]
Some small loader refinements and refactoring
https://bugs.webkit.org/show_bug.cgi?id=131541

Reviewed by Alexey Proskuryakov.

Cut down on use of ResourceLoader::handle, which always returns null when
using a network process. Also streamlined conditionals and did some other
small refactoring.

* loader/DocumentLoader.h: Use references rather than pointers for SchedulePair.

* loader/ResourceLoader.cpp:
(WebCore::ResourceLoader::didChangePriority): Use m_handle instead of handle(),
since we'd like to delete handle() entirely soon.
(WebCore::ResourceLoader::didReceiveAuthenticationChallenge): Ditto.
(WebCore::ResourceLoader::schedule): Added.
(WebCore::ResourceLoader::unschedule): Ditto.
* loader/ResourceLoader.h: Rearranged header to eliminate nested conditionals.
Added schedule and unschedule functions for Mac.

* loader/mac/DocumentLoaderMac.cpp:
(WebCore::scheduleAll): Changed to take a reference and call ResourceLoader::schedule.
(WebCore::unscheduleAll): Ditto.
(WebCore::DocumentLoader::schedule): Ditto.
(WebCore::DocumentLoader::unschedule): Ditto.

* page/mac/PageMac.cpp:
(WebCore::Page::addSchedulePair): Pass a reference rather than a pointer to schedule.
(WebCore::Page::removeSchedulePair): Ditto.

* platform/network/ResourceHandle.h: Did a bit of reformatting and reorganizing of
conditionals.

* platform/network/ResourceHandleInternal.h: Removed a tiny bit of unneeded declaration.

* platform/network/mac/ResourceHandleMac.mm:
(WebCore::ResourceHandle::schedule): Updated to take a reference rather than a pointer.
(WebCore::ResourceHandle::unschedule): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167183 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoDFG::FixupPhase should insert conversion nodes after the rest of fixup so that we...
fpizlo@apple.com [Sat, 12 Apr 2014 18:22:27 +0000 (18:22 +0000)]
DFG::FixupPhase should insert conversion nodes after the rest of fixup so that we know how the types settled
https://bugs.webkit.org/show_bug.cgi?id=131424

Reviewed by Geoffrey Garen.

This defers type conversion injection until we've decided on types. This makes the
process of deciding types a bit more flexible - for example we can naturally fixpoint
and change our minds. Only when things are settled do we actually insert conversions.

This is a necessary prerequisite for keeping double, int52, and JSValue data flow
separate. A SetLocal/GetLocal will appear to be JSValue until we fixpoint and realize
that there are typed uses. If we were eagerly inserting type conversions then we would
first insert a to/from-JSValue conversion in some cases only to then replace it by
the other conversions. It's probably trivial to remove those redundant conversions later
but I think it's better if we don't insert them to begin with.

* bytecode/CodeOrigin.h:
(JSC::CodeOrigin::operator!):
* dfg/DFGFixupPhase.cpp:
(JSC::DFG::FixupPhase::run):
(JSC::DFG::FixupPhase::fixupBlock):
(JSC::DFG::FixupPhase::fixupNode):
(JSC::DFG::FixupPhase::fixupSetLocalsInBlock):
(JSC::DFG::FixupPhase::fixEdge):
(JSC::DFG::FixupPhase::fixIntEdge):
(JSC::DFG::FixupPhase::injectTypeConversionsInBlock):
(JSC::DFG::FixupPhase::injectTypeConversionsForEdge):
(JSC::DFG::FixupPhase::addRequiredPhantom):
(JSC::DFG::FixupPhase::addPhantomsIfNecessary):
(JSC::DFG::FixupPhase::clearPhantomsAtEnd):
(JSC::DFG::FixupPhase::observeUntypedEdge): Deleted.
(JSC::DFG::FixupPhase::fixupUntypedSetLocalsInBlock): Deleted.
(JSC::DFG::FixupPhase::injectInt32ToDoubleNode): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167182 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoMake NodeList.length inline-cacheable by JSC.
akling@apple.com [Sat, 12 Apr 2014 05:59:34 +0000 (05:59 +0000)]
Make NodeList.length inline-cacheable by JSC.
<https://webkit.org/b/131579>

For objects with custom index or name getters, we have to make sure
that builtin properties take precedence. We do this by scanning the
ancestor chain for a suitable property slot before moving on to
named items.

With this patch, we now mark such builtins as cacheable. This is
safe since the whole point of doing this before processing named
items is to ensure the same slot is returned consistently.

Reviewed by Benjamin Poulain.

* bindings/scripts/CodeGeneratorJS.pm:
(GenerateGetOwnPropertySlotBody):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167181 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[Mac] Crashes when copying or pasting huge images
ap@apple.com [Sat, 12 Apr 2014 03:09:38 +0000 (03:09 +0000)]
[Mac] Crashes when copying or pasting huge images
https://bugs.webkit.org/show_bug.cgi?id=131576
<rdar://problem/12131833>
<rdar://problem/14427398>

Reviewed by Darin Adler.

Added a few null checks for SharedMemory::create() return value in pasteboard code.
Error handling feels a bit sketchy, but
- I'm not sure what it should look like ideally;
- it matches the kind of error handling we already have in these functions;
- it appears to work reasonably well in practice. We get empty content, which
is not nice, but not particularly problematic either. When copying an animated GIF,
we also get the GIF in RTFD flavor, so even pasting into NSTextViews works!

* Platform/mac/SharedMemoryMac.cpp:
(WebKit::SharedMemory::createFromVMBuffer):
* UIProcess/mac/WebContextMac.mm:
(WebKit::WebContext::getPasteboardBufferForType):
(WebKit::WebContext::readBufferFromPasteboard):
* WebProcess/WebCoreSupport/WebPlatformStrategies.cpp:
(WebKit::WebPlatformStrategies::setBufferForType):

* WebProcess/WebCoreSupport/mac/WebDragClientMac.mm: (WebKit::WebDragClient::declareAndWriteDragImage):
Also renamed some variables to prevent name collisions with with nested scope.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167180 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoUnreviewed build fix on the EFL port after r167152
ryuan.choi@samsung.com [Sat, 12 Apr 2014 02:49:02 +0000 (02:49 +0000)]
Unreviewed build fix on the EFL port after r167152

* WebProcess/WebPage/PageOverlay.cpp:
(WebKit::PageOverlay::bounds):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167179 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[WK2] WebProcess crashes, when closing window after opening page by means of context...
commit-queue@webkit.org [Sat, 12 Apr 2014 01:56:06 +0000 (01:56 +0000)]
[WK2] WebProcess crashes, when closing window after opening page by means of context menu
https://bugs.webkit.org/show_bug.cgi?id=131439

Patch by Hyowon Kim <hw1008.kim@samsung.com> on 2014-04-11
Reviewed by Tim Horton.

CoordinatedGraphicsLayer calls notifyFlushRequired() through its client when destroyed.
When PageOverlayController::notifyFlushRequired() is called, the DrawingArea could be null
because it's destoryed in WebPage::close().

* WebProcess/WebPage/PageOverlayController.cpp:
(WebKit::PageOverlayController::notifyFlushRequired): Add null check of m_webPage->drawingArea().

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167178 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoLots of compositing test failures after r167152
timothy_horton@apple.com [Sat, 12 Apr 2014 01:24:37 +0000 (01:24 +0000)]
Lots of compositing test failures after r167152
https://bugs.webkit.org/show_bug.cgi?id=131574

Reviewed by Simon Fraser.

* WebProcess/WebPage/PageOverlayController.h:
Skip page overlay layers in layer tree dumps, for consistency between platforms.

* platform/graphics/GraphicsLayer.cpp:
(WebCore::GraphicsLayer::dumpProperties):
We were asking the parent's client whether the child should be dumped,
but really the child's client is the one who should decide.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167177 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoViewportConfiguration::layoutWidth() compute the width incorrectly when initial-scale...
benjamin@webkit.org [Sat, 12 Apr 2014 00:43:37 +0000 (00:43 +0000)]
ViewportConfiguration::layoutWidth() compute the width incorrectly when initial-scale+width do not fit in view
https://bugs.webkit.org/show_bug.cgi?id=131575

Patch by Benjamin Poulain <bpoulain@apple.com> on 2014-04-11
Reviewed by Enrica Casucci.

I made a mistake when writing the new viewport code: ViewportConfiguration::layoutWidth() uses m_contentSize in one place.
That make no sense, contentSize depends on the layout width.

* page/ViewportConfiguration.cpp:
(WebCore::ViewportConfiguration::layoutWidth):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167176 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoRemove "numeric index getter" stuff from bindings code generator.
akling@apple.com [Sat, 12 Apr 2014 00:26:02 +0000 (00:26 +0000)]
Remove "numeric index getter" stuff from bindings code generator.
<https://webkit.org/b/131565>

This was used for an earlier generation of typed arrays. Now that JSC
has native typed array support, we're not using this functionality and
can just remove it.

Reviewed by Geoffrey Garen.

* bindings/scripts/CodeGeneratorJS.pm:
(GenerateGetOwnPropertySlotBody):
(HasComplexGetOwnProperty):
(InterfaceRequiresAttributesOnInstance):
(InstanceOverridesGetOwnPropertySlot):
(GenerateHeader):
(GenerateImplementation):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167175 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoWeb Replay: code generator should consider enclosing class when computing duplicate...
burg@cs.washington.edu [Sat, 12 Apr 2014 00:25:06 +0000 (00:25 +0000)]
Web Replay: code generator should consider enclosing class when computing duplicate type names
https://bugs.webkit.org/show_bug.cgi?id=131554

Reviewed by Timothy Hatcher.

We need to prepend an enum's enclosing class, if any, so that multiple enums with the same name
can coexist without triggering a "duplicate types" error. Now, such enums must be referenced
by the enclosing class and enum name.

Added tests for the new syntax, and rebaselined one test to reflect a previous patch's change.

* replay/scripts/CodeGeneratorReplayInputs.py:
(Type.type_name): Prepend the enclosing class name.
(Type.type_name.is):
* replay/scripts/tests/expected/fail-on-duplicate-enum-type.json-error: Added.
* replay/scripts/tests/expected/generate-enums-with-same-base-name.json-TestReplayInputs.cpp: Added.
* replay/scripts/tests/expected/generate-enums-with-same-base-name.json-TestReplayInputs.h: Added.
* replay/scripts/tests/expected/generate-input-with-vector-members.json-TestReplayInputs.h: Rebaseline.
* replay/scripts/tests/fail-on-duplicate-enum-type.json: Added.
* replay/scripts/tests/generate-enums-with-same-base-name.json: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167174 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoWeb Replay: consolidate decoding macros used in SerializationMethods
burg@cs.washington.edu [Sat, 12 Apr 2014 00:19:24 +0000 (00:19 +0000)]
Web Replay: consolidate decoding macros used in SerializationMethods
https://bugs.webkit.org/show_bug.cgi?id=131564

Reviewed by Timothy Hatcher.

We can use EncodingTraits<T>::DecodedType to create a local variable
with the appropriate decoded type, rather than special casing.

* replay/SerializationMethods.cpp:
(JSC::EncodingTraits<NondeterministicInputBase>::decodeValue):
(JSC::EncodingTraits<KeypressCommand>::decodeValue):
(JSC::EncodingTraits<PlatformKeyboardEvent>::decodeValue):
(JSC::EncodingTraits<PlatformMouseEvent>::decodeValue):
(JSC::EncodingTraits<PlatformWheelEvent>::decodeValue):
(JSC::EncodingTraits<PluginData>::decodeValue):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167173 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoNeed WK2 API to disable rubber-banding
bdakin@apple.com [Fri, 11 Apr 2014 23:51:36 +0000 (23:51 +0000)]
Need WK2 API to disable rubber-banding
https://bugs.webkit.org/show_bug.cgi?id=131567

Reviewed by Simon Fraser.

Source/WebCore:

Page will now store a vertical and horizontal ScrollElasticity in case the API has
been called before the FrameView even exists.

New Page functions.
* WebCore.exp.in:

Use Page’s cached ScrollElasticity.
* page/FrameView.cpp:
(WebCore::FrameView::FrameView):

If there is already a FrameView, then assign it the new ScrollElasticity.
* page/Page.cpp:
(WebCore::Page::Page):
(WebCore::Page::setVerticalScrollElasticity):
(WebCore::Page::setHorizontalScrollElasticity):
* page/Page.h:
(WebCore::Page::verticalScrollElasticity):
(WebCore::Page::horizontalScrollElasticity):

Source/WebKit2:

New API allows the client to enable/disable horizontal or vertical rubber-banding.
By default, rubber-banding is enabled.
* UIProcess/API/C/WKPage.cpp:
(WKPageVerticalRubberBandingIsEnabled):
(WKPageSetEnableVerticalRubberBanding):
(WKPageHorizontalRubberBandingIsEnabled):
(WKPageSetEnableHorizontalRubberBanding):
* UIProcess/API/C/WKPage.h:
* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::WebPageProxy):
(WebKit::WebPageProxy::setEnableVerticalRubberBanding):
(WebKit::WebPageProxy::verticalRubberBandingIsEnabled):
(WebKit::WebPageProxy::setEnableHorizontalRubberBanding):
(WebKit::WebPageProxy::horizontalRubberBandingIsEnabled):
* UIProcess/WebPageProxy.h:
* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::setEnableVerticalRubberBanding):
(WebKit::WebPage::setEnableHorizontalRubberBanding):
* WebProcess/WebPage/WebPage.h:
* WebProcess/WebPage/WebPage.messages.in:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167172 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoFix the Mac build.
timothy_horton@apple.com [Fri, 11 Apr 2014 23:50:03 +0000 (23:50 +0000)]
Fix the Mac build.

* WebProcess/WebPage/FindController.cpp:
(WebKit::FindController::updateFindIndicator):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167171 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years agoREGRESSION(167145): Many media tests fail
mrobinson@webkit.org [Fri, 11 Apr 2014 23:49:32 +0000 (23:49 +0000)]
REGRESSION(167145): Many media tests fail
https://bugs.webkit.org/show_bug.cgi?id=131569

Reviewed by Brent Fulgham.

* PlatformGTK.cmake: Add the localized strings file to the list of user agent scripts.
* platform/gtk/RenderThemeGtk.cpp:
(WebCore::RenderThemeGtk::mediaControlsScript): Include the localized strings file in the script body.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167170 268f45cc-cd09-0410-ab3c-d52691b4dbfc

7 years ago[iOS WebKit2] Find-in-page indicator
timothy_horton@apple.com [Fri, 11 Apr 2014 23:31:27 +0000 (23:31 +0000)]
[iOS WebKit2] Find-in-page indicator
https://bugs.webkit.org/show_bug.cgi?id=131510
<rdar://problem/16547777>

Reviewed by Simon Fraser and Enrica Casucci.

* WebKit2.xcodeproj/project.pbxproj:
* WebProcess/WebPage/FindController.cpp:
(WebKit::FindController::updateFindUIAfterPageScroll):
WebKit2 will clear the selection if we reveal it while not enabling selection painting.
So, avoid revealing it.

(WebKit::FindController::findString):
(WebKit::FindController::hideFindIndicator):
(WebKit::FindController::willFindString):
(WebKit::FindController::didFailToFindString):
(WebKit::FindController::didHideFindIndicator):
#if out the cross-platform find indicator code on iOS; it will be replaced
by the code in FindControllerIOS.

* WebProcess/WebPage/FindController.h:

* WebProcess/WebPage/ios/FindControllerIOS.mm: Added.
(highlightColor):
(WebKit::FindIndicatorOverlayClientIOS::drawRect):
Paint the selection (with black text forced on), and the yellow rounded rect
into the indicator page overlay.

(WebKit::FindController::updateFindIndicator):
Create or update a small document-relative page overlay in the selection rect.

(WebKit::FindController::hideFindIndicator):
(WebKit::FindController::willFindString):
(WebKit::FindController::didFailToFindString):
(WebKit::FindController::didHideFindIndicator):
Add willFindString, didFailToFindString, and didHideFindIndicator functions.
They do nothing, except on iOS where they are used to prevent selection change
messages from being dispatched while using find-in-page, and to force
WebCore to compute a selection rect (via updateAppearance) despite the selection not being painted.

* WebProcess/WebPage/ios/FindIndicatorOverlayClientIOS.h:
(WebKit::FindIndicatorOverlayClientIOS::setFrame):

* editing/Editor.cpp:
(WebCore::Editor::findString):
* editing/FindOptions.h:
Add a find option which prevents WebCore from revealing the selection
after selecting a successful find match.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@167169 268f45cc-cd09-0410-ab3c-d52691b4dbfc