WebKit-https.git
20 months agoAdd pop-up policy support in website policies.
commit-queue@webkit.org [Tue, 16 Jan 2018 19:12:43 +0000 (19:12 +0000)]
Add pop-up policy support in website policies.
https://bugs.webkit.org/show_bug.cgi?id=181544.
rdar://problem/30521400.

Patch by Zach Li <zachli@apple.com> on 2018-01-16
Reviewed by Alex Christensen.

Source/WebCore:

* loader/DocumentLoader.h:
Introduce pop-up policy getter and setter. Initialize
the policy as Default.
(WebCore::DocumentLoader::popUpPolicy const):
(WebCore::DocumentLoader::setPopUpPolicy):
* page/DOMWindow.cpp:
(WebCore::DOMWindow::allowPopUp):
Pop-up policy specified on a per-page basis holds
precedence over the global policy. If no pop-up policy
is specified during navigation, global policy is used.

Source/WebKit:

* Shared/WebsitePoliciesData.cpp: Encode and decode
pop-up policy.
(WebKit::WebsitePoliciesData::encode const):
(WebKit::WebsitePoliciesData::decode):
(WebKit::WebsitePoliciesData::applyToDocumentLoader):
Apply the pop-up policy from website policies to the
document loader.

* Shared/WebsitePoliciesData.h:

* Shared/WebsitePopUpPolicy.h:
Added for declaring enum WebsitePopUpPolicy.

* UIProcess/API/APIWforebsitePolicies.cpp:
Include pop-up policy in website policies.
(API::WebsitePolicies::WebsitePolicies):
(API::WebsitePolicies::data):
* UIProcess/API/APIWebsitePolicies.h:

* UIProcess/API/C/WKWebsitePolicies.cpp:
Added C API to get and set pop-up policy on
website policies.
(WKWebsitePoliciesGetPopUpPolicy):
(WKWebsitePoliciesSetPopUpPolicy):
* UIProcess/API/C/WKWebsitePolicies.h:

* UIProcess/API/Cocoa/_WKWebsitePolicies.h:
Added Obj-C API to get and set pop-up policy
on website policies.
* UIProcess/API/Cocoa/_WKWebsitePolicies.mm:
(-[_WKWebsitePolicies setPopUpPolicy:]):
(-[_WKWebsitePolicies popUpPolicy]):

* WebKit.xcodeproj/project.pbxproj:
Added WebsitePopUpPolicy.h.

Tools:

* TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
Added pop-up-check.html.

* TestWebKitAPI/Tests/WebKit/pop-up-check.html: Added.

* TestWebKitAPI/Tests/WebKitCocoa/WebsitePolicies.mm:
(-[PopUpPoliciesDelegate webView:decidePolicyForNavigationAction:decisionHandler:]):
(-[PopUpPoliciesDelegate _webView:decidePolicyForNavigationAction:decisionHandler:]):
(-[PopUpPoliciesDelegate webView:createWebViewWithConfiguration:forNavigationAction:windowFeatures:]):
(TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226984 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months ago[Win] JSC compile error in runtime/DateConversion.cpp.
pvollan@apple.com [Tue, 16 Jan 2018 19:09:32 +0000 (19:09 +0000)]
[Win] JSC compile error in runtime/DateConversion.cpp.
https://bugs.webkit.org/show_bug.cgi?id=181690
<rdar://problem/36480409>

Reviewed by Brent Fulgham.

Add 'text/win' folder to list of forwarding headers directories.

* wtf/PlatformWin.cmake:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226983 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months agoCrash playing audio-only HLS stream via hls.js (MSE)
jer.noble@apple.com [Tue, 16 Jan 2018 19:04:22 +0000 (19:04 +0000)]
Crash playing audio-only HLS stream via hls.js (MSE)
https://bugs.webkit.org/show_bug.cgi?id=181691
<rdar://problem/32967295>

Reviewed by Eric Carlson.

Add a weak-link check to the block called by -[AVSampleBufferDisplayLayer requestMediaDataWhenReadyOnQueue:usingBlock:].

* platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:
(WebCore::SourceBufferPrivateAVFObjC::trackDidChangeEnabled):
(WebCore::SourceBufferPrivateAVFObjC::notifyClientWhenReadyForMoreSamples):
(WebCore::SourceBufferPrivateAVFObjC::setVideoLayer):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226982 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months agoRename applyHorizontalScale/applyVerticalScale in SVG filters, and related cleanup
simon.fraser@apple.com [Tue, 16 Jan 2018 18:59:09 +0000 (18:59 +0000)]
Rename applyHorizontalScale/applyVerticalScale in SVG filters, and related cleanup
https://bugs.webkit.org/show_bug.cgi?id=181684

Reviewed by Alex Christensen.

Rename the confusing applyHorizontalScale/applyVerticalScale to scaledByFilterResolution(),
and have it take and return a FloatSize. Change callers to do math in terms of FloatSizes.

Add inflate(size) to each of the rect classes.

* platform/graphics/FloatRect.h:
(WebCore::FloatRect::inflate):
* platform/graphics/IntRect.h:
(WebCore::IntRect::inflate):
* platform/graphics/LayoutRect.h:
(WebCore::LayoutRect::inflate):
* platform/graphics/filters/FEDisplacementMap.cpp:
(WebCore::FEDisplacementMap::platformApplySoftware):
* platform/graphics/filters/FEDropShadow.cpp:
(WebCore::FEDropShadow::determineAbsolutePaintRect):
(WebCore::FEDropShadow::platformApplySoftware):
* platform/graphics/filters/FEGaussianBlur.cpp:
(WebCore::FEGaussianBlur::calculateUnscaledKernelSize):
(WebCore::FEGaussianBlur::calculateKernelSize):
(WebCore::FEGaussianBlur::determineAbsolutePaintRect):
(WebCore::FEGaussianBlur::platformApplySoftware):
* platform/graphics/filters/FEGaussianBlur.h:
* platform/graphics/filters/FEMorphology.cpp:
(WebCore::FEMorphology::determineAbsolutePaintRect):
(WebCore::FEMorphology::platformApplySoftware):
* platform/graphics/filters/FEOffset.cpp:
(WebCore::FEOffset::determineAbsolutePaintRect):
(WebCore::FEOffset::platformApplySoftware):
* platform/graphics/filters/Filter.h:
(WebCore::Filter::setSourceImage):
(WebCore::Filter::scaledByFilterResolution const):
(WebCore::Filter::applyHorizontalScale const): Deleted.
(WebCore::Filter::applyVerticalScale const): Deleted.
* platform/graphics/filters/FilterOperations.cpp:
(WebCore::outsetSizeForBlur):
* rendering/FilterEffectRenderer.h:
* svg/graphics/filters/SVGFilter.cpp:
(WebCore::SVGFilter::scaledByFilterResolution const):
(WebCore::SVGFilter::applyHorizontalScale const): Deleted.
(WebCore::SVGFilter::applyVerticalScale const): Deleted.
* svg/graphics/filters/SVGFilter.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226981 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months agoFix Mac CMake build after r222900
achristensen@apple.com [Tue, 16 Jan 2018 18:58:47 +0000 (18:58 +0000)]
Fix Mac CMake build after r222900

* MallocBench/MallocBench/CMakeLists.txt:
The mbmalloc target doesn't exist if we're not in developer mode, so only use it if we are.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226980 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months ago[CMake][Mac] Fix the build errors
achristensen@apple.com [Tue, 16 Jan 2018 18:55:09 +0000 (18:55 +0000)]
[CMake][Mac] Fix the build errors
https://bugs.webkit.org/show_bug.cgi?id=181665

Patch by Fujii Hironori <Hironori.Fujii@sony.com> on 2018-01-16
Reviewed by Alex Christensen.

Source/WebCore:

No new tests (No behavior change)

* CMakeLists.txt: Added Modules/paymentrequest/MerchantValidationEvent.idl to compile.
* PlatformMac.cmake: Added Modules/paymentrequest/MerchantValidationEvent.cpp to compile.
Added workers/service/context and Modules/applicationmanifest as forwarding header paths.

Source/WebKit:

* CMakeLists.txt: Added TouchBarMenuData.cpp and TouchBarMenuItemData.cpp to compile.
* PlatformMac.cmake: Added _WKApplicationManifest.mm, WKInspectorWKWebView.mm and WKInspectorWindow.mm to compile.
* UIProcess/API/Cocoa/WKProcessPool.mm:
(-[WKProcessPool _webPageContentProcessCount]): Call WebProcessPool::serviceWorkerProxy() only if ENABLE(SERVICE_WORKER).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226979 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months ago[Attachment Support] Provide the `src` of an attachment to the UI delegate when an...
wenson_hsieh@apple.com [Tue, 16 Jan 2018 18:46:04 +0000 (18:46 +0000)]
[Attachment Support] Provide the `src` of an attachment to the UI delegate when an attachment is inserted
https://bugs.webkit.org/show_bug.cgi?id=181638
<rdar://problem/36508702>

Reviewed by Dan Bernstein.

Source/WebCore:

Adjust the `didInsertAttachment` codepath to additionally propagate the attachment element's `src`.
Additionally, fix an issue with insertion and removal client notifications wherein the client can receive
insertion calls without corresponding removal calls, or vice versa. This is an existing issue, but matters more
now because we actually need to access the attachment element for its `src` when propagating changes to the
client. See below for details.

Test: WKAttachmentTests.AttachmentUpdatesWhenInsertingRichMarkup

* dom/Document.h:
(WebCore::Document::attachmentElementsByIdentifier const):
* editing/Editor.cpp:
(WebCore::Editor::notifyClientOfAttachmentUpdates):
* page/EditorClient.h:
(WebCore::EditorClient::didInsertAttachment):
* page/Frame.cpp:
(WebCore::Frame::setDocument):

When a Frame's document changes, inform the client that the attachments in the previous document are going away.
For each attachment currently connected to the document, we have either (1) already informed the client that it
was inserted, or (2) the attachment is pending an insertion call to the client. If (1) is the case, then we'll
tell the client that the attachment is removed, which will balance out the earlier insertion call. If (2) is the
case, then we'll remove the previously inserted attachment identifier from the set of attachment identifiers
pending insertion, and the client won't be informed of insertions or removals.

Source/WebKit:

Add a `source` parameter to the `didInsertAttachment` codepath for notifying WebKit2 clients when attachment
elements are inserted into the document.

* UIProcess/API/Cocoa/WKUIDelegatePrivate.h:
* UIProcess/API/Cocoa/WKWebView.mm:
(-[WKWebView _didInsertAttachment:withSource:]):
(-[WKWebView _didInsertAttachment:]): Deleted.
* UIProcess/API/Cocoa/WKWebViewInternal.h:
* UIProcess/Cocoa/PageClientImplCocoa.h:
* UIProcess/Cocoa/PageClientImplCocoa.mm:
(WebKit::PageClientImplCocoa::didInsertAttachment):
* UIProcess/PageClient.h:
(WebKit::PageClient::didInsertAttachment):
* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::didInsertAttachment):
* UIProcess/WebPageProxy.h:
* UIProcess/WebPageProxy.messages.in:
* WebProcess/WebCoreSupport/WebEditorClient.cpp:
(WebKit::WebEditorClient::didInsertAttachment):
* WebProcess/WebCoreSupport/WebEditorClient.h:

Tools:

Tweak an existing API test to check that the `src` of an attachment element inserted via script matches the
`source` provided to the UI delegate via -[WKUIDelegate _webView:didInsertAttachment:withSource:].

* TestWebKitAPI/Tests/WebKitCocoa/WKAttachmentTests.mm:
(-[AttachmentUpdateObserver init]):
(-[AttachmentUpdateObserver sourceForIdentifier:]):
(-[AttachmentUpdateObserver _webView:didInsertAttachment:withSource:]):
(TestWebKitAPI::ObserveAttachmentUpdatesForScope::expectSourceForIdentifier):
(TestWebKitAPI::TEST):
(-[AttachmentUpdateObserver _webView:didInsertAttachment:]): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226977 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months agoUse traits for animation timing functions
commit-queue@webkit.org [Tue, 16 Jan 2018 18:45:45 +0000 (18:45 +0000)]
Use traits for animation timing functions
https://bugs.webkit.org/show_bug.cgi?id=181651

Patch by Antoine Quint <graouts@apple.com> on 2018-01-16
Reviewed by Dean Jackson.

Cleaning up Dean's previous patch as suggested by Darin's post-commit review comments. The
downcast function can match const automatically and it's a better style to put the * inside
the downcast call rather than outside.

* css/CSSComputedStyleDeclaration.cpp:
(WebCore::createTimingFunctionValue):
* platform/animation/TimingFunction.cpp:
(WebCore::operator<<):
(WebCore::TimingFunction::transformTime const):
* platform/animation/TimingFunction.h:
* platform/graphics/ca/cocoa/PlatformCAAnimationCocoa.mm:
(WebCore::toCAMediaTimingFunction):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226976 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months ago[GStreamer] Live streaming cleanups
philn@webkit.org [Tue, 16 Jan 2018 16:01:20 +0000 (16:01 +0000)]
[GStreamer] Live streaming cleanups
https://bugs.webkit.org/show_bug.cgi?id=181672

Reviewed by Michael Catanzaro.

* platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.cpp:
(WebCore::MediaPlayerPrivateGStreamer::handleMessage): Switch to
streaming code path when no content-length was reported by the
http source element.
(WebCore::MediaPlayerPrivateGStreamer::totalBytes const): Return early when streaming.
* platform/graphics/gstreamer/WebKitWebSourceGStreamer.cpp:
(CachedResourceStreamingClient::responseReceived): Emit headers as
an element message, like souphttpsrc.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226974 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months ago[GStreamer] misc fixes and cleanups
philn@webkit.org [Tue, 16 Jan 2018 15:59:59 +0000 (15:59 +0000)]
[GStreamer] misc fixes and cleanups
https://bugs.webkit.org/show_bug.cgi?id=181647

Reviewed by Michael Catanzaro.

* platform/graphics/MediaPlayer.cpp:
(WebCore::convertEnumerationToString): New utility function to convert preload enum to string.
* platform/graphics/MediaPlayerEnums.h: Ditto.
* platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.cpp:
(WebCore::MediaPlayerPrivateGStreamer::load): Debug tweak
(WebCore::MediaPlayerPrivateGStreamer::prepareToPlay): Ditto
(WebCore::MediaPlayerPrivateGStreamer::play): Ditto
(WebCore::MediaPlayerPrivateGStreamer::paused const): Ditto
(WebCore::MediaPlayerPrivateGStreamer::processBufferingStats): Prevent useless state update.
(WebCore::MediaPlayerPrivateGStreamer::fillTimerFired): Ditto.
(WebCore::MediaPlayerPrivateGStreamer::updateStates): Debug tweak.
(WebCore::MediaPlayerPrivateGStreamer::setDownloadBuffering): Ditto.
* platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.h:
Remove useless handlesSyncMessage method.
* platform/graphics/gstreamer/MediaPlayerPrivateGStreamerBase.cpp:
(WebCore::MediaPlayerPrivateGStreamerBase::setMuted): Prevent useless state update.
(WebCore::MediaPlayerPrivateGStreamerBase::muted const): Debug tweak.
(WebCore::MediaPlayerPrivateGStreamerBase::setStreamVolumeElement): Ditto.
* platform/graphics/gstreamer/WebKitWebSourceGStreamer.cpp:
(webkit_web_src_init): Enable keepAlive by default like in souphttpsrc.
(webKitWebSrcStart): Debug tweak.
(webKitWebSrcQueryWithParent): Ditto.
(webKitWebSrcNeedData): Ditto.
(CachedResourceStreamingClient::responseReceived): Change appsrc
stream-type when we wan't seek. Also update caps like souphttpsrc
does.
* platform/graphics/gstreamer/WebKitWebSourceGStreamer.h: Add forward declaration of MediaPlayer.
* platform/graphics/gstreamer/mse/MediaSourceClientGStreamerMSE.cpp:
(WebCore::MediaSourceClientGStreamerMSE::append): Fix compilation warning.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226973 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months agombmalloc should only be built in developer mode
mcatanzaro@igalia.com [Tue, 16 Jan 2018 15:34:07 +0000 (15:34 +0000)]
mbmalloc should only be built in developer mode
https://bugs.webkit.org/show_bug.cgi?id=181654

Reviewed by Carlos Garcia Campos.

* CMakeLists.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226972 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months agoTest programs should only be built in developer mode
mcatanzaro@igalia.com [Tue, 16 Jan 2018 15:33:22 +0000 (15:33 +0000)]
Test programs should only be built in developer mode
https://bugs.webkit.org/show_bug.cgi?id=181653

Reviewed by Carlos Garcia Campos.

Build test programs only in developer mode, and fix code style.

* shell/CMakeLists.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226971 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months agoUnreviewed. Fix GTK unit tests execution in the bots after r226967.
carlosgc@webkit.org [Tue, 16 Jan 2018 14:18:40 +0000 (14:18 +0000)]
Unreviewed. Fix GTK unit tests execution in the bots after r226967.

I forgot to pass the environment to the GLibTestRunner and Xvfb was not used in the bots.

* glib/api_test_runner.py:
(TestRunner._run_test_glib): Pass env to GLibTestRunner.
* glib/glib_test_runner.py:
(GLibTestRunner.run): Use given env.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226970 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months ago[GTK][WPE] Stop passing --verbose to run-api-tests from the bots
carlosgc@webkit.org [Tue, 16 Jan 2018 12:10:37 +0000 (12:10 +0000)]
[GTK][WPE] Stop passing --verbose to run-api-tests from the bots
https://bugs.webkit.org/show_bug.cgi?id=181676

Reviewed by Carlos Alberto Lopez Perez.

The option was removed in r226967.

* BuildSlaveSupport/build.webkit.org-config/steps.py:
(RunGtkAPITests):
(RunWPEAPITests):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226969 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months agoUnreviewed. Temporarily bring back --verbose option of api_test_runner.py.
carlosgc@webkit.org [Tue, 16 Jan 2018 11:38:59 +0000 (11:38 +0000)]
Unreviewed. Temporarily bring back --verbose option of api_test_runner.py.

The bots are still passing that option and the step is failing because it was removed in r226967.

* glib/api_test_runner.py:
(add_options):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226968 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months ago[GTK][WPE] Improve the way glib tests are run
commit-queue@webkit.org [Tue, 16 Jan 2018 10:42:49 +0000 (10:42 +0000)]
[GTK][WPE] Improve the way glib tests are run
https://bugs.webkit.org/show_bug.cgi?id=181674

Patch by Carlos Garcia Campos <cgarcia@igalia.com> on 2018-01-16
Reviewed by Žan Doberšek.

Stop using gtester external program and use our own tester from python directly. This way we no longer need to
parse the tests output to get the results which is causing problems in WPE bot. We can now differentiate between
tests failing due to an expected assert in the test and unexpected crashes.
This also fixes a bug in previous code where we failed to properly detect tests timing out, because gtester was
not showing the subtest name in stdout in case of timeouts.
I've lowered the default timeout from 10 to 5, since we are now properly handling the timeout for every test
case. I've also removed the verbose option, since it was only used by gtester and we now always show the result
of every test case.

* glib/api_test_runner.py:
(TestRunner._run_test_glib): Use GLibTestRunner.
(TestRunner._run_google_test): Wrote tests timing out to stdout too.
(add_options):
* glib/glib_test_runner.py: Added.
(TestTimeout):
(Message):
(Message.__init__):
(Message.create):
(Message.create.read_unsigned):
(Message.create.read_double):
(Message.create.read_string):
(GLibTestRunner):
(GLibTestRunner.__init__):
(GLibTestRunner._process_data):
(GLibTestRunner._process_message):
(GLibTestRunner._read_from_pipe):
(GLibTestRunner._read_from_stderr):
(GLibTestRunner._start_timeout):
(GLibTestRunner._start_timeout._alarm_handler):
(GLibTestRunner._stop_timeout):
(GLibTestRunner._subtest_start):
(GLibTestRunner._subtest_message):
(GLibTestRunner._subtest_stderr):
(GLibTestRunner._subtest_end):
(GLibTestRunner.run):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226967 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months ago[EME] Add layout test for the MSE playback of ClearKey encrypted audio and video
commit-queue@webkit.org [Tue, 16 Jan 2018 08:16:32 +0000 (08:16 +0000)]
[EME] Add layout test for the MSE playback of ClearKey encrypted audio and video
https://bugs.webkit.org/show_bug.cgi?id=181594

Patch by Yacine Bandou <yacine.bandou_ext@softathome.com> on 2018-01-16
Reviewed by Xabier Rodriguez-Calvar.

Add new layout tests in order to test the playback of ClearKey encrypted content:
clearKey-cenc-audio-playback-mse : Tests the MSE playback of ClearKey encrypted audio track.
clearKey-cenc-video-playback-mse : Tests the MSE playback of ClearKey encrypted video track.

* media/content/encrypted/segments/AudioClearKeyCenc-seg-0.mp4: Added.
* media/content/encrypted/segments/AudioClearKeyCenc-seg-1.mp4: Added.
* media/content/encrypted/segments/AudioClearKeyCenc-seg-2.mp4: Added.
* media/content/encrypted/segments/AudioClearKeyCenc-seg-3.mp4: Added.
    These are a simple encrypted audio fragments used in MSE playback.
    They have been fragmented and encrypted with BENTO4. https://www.bento4.com/
    Use full-sample encryption because they are not NAL Structured.
* media/content/encrypted/segments/VideoClearKeyCenc-seg-1.mp4: Added.
* media/content/encrypted/segments/VideoClearKeyCenc-seg-2.mp4: Added.
* media/content/encrypted/segments/VideoClearKeyCenc-seg-3.mp4: Added.
    These are a simple encrypted video fragments used in MSE playback.
    They have been fragmented and encrypted with BENTO4. https://www.bento4.com/
    Use sub-sample encryption because they are NAL Structured.
* media/encrypted-media/clearKey/clearKey-cenc-audio-playback-mse-expected.txt: Added.
* media/encrypted-media/clearKey/clearKey-cenc-audio-playback-mse.html: Added.
* media/encrypted-media/clearKey/clearKey-cenc-video-playback-mse-expected.txt: Added.
* media/encrypted-media/clearKey/clearKey-cenc-video-playback-mse.html: Added.
* media/encrypted-media/medias-enc.js:
* platform/gtk/TestExpectations:
* platform/mac/TestExpectations:
* platform/wpe/TestExpectations:
    The test clearKey-cenc-audio-playback-mse crashes because the clearKey decryptor plugin doesn't
    support yet the Full-sample encryption, see bug 180080.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226966 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months ago[WPE] Two clearkey tests failing since r226621
commit-queue@webkit.org [Tue, 16 Jan 2018 08:13:04 +0000 (08:13 +0000)]
[WPE] Two clearkey tests failing since r226621
https://bugs.webkit.org/show_bug.cgi?id=181532

Patch by Yacine Bandou <yacine.bandou_ext@softathome.com> on 2018-01-16
Reviewed by Xabier Rodriguez-Calvar.

Source/WebCore:

Whith a fake initData, we can have a pssh size nul, thus we should check it.
We saw this issue in the subtest "initData longer than 64Kb characters" in
the clearkey-generate-request-disallowed-input layout test.

* platform/encryptedmedia/clearkey/CDMClearKey.cpp:
(WebCore::extractKeyidsLocationFromCencInitData):

LayoutTests:

* platform/wpe/imported/w3c/web-platform-tests/encrypted-media/clearkey-generate-request-disallowed-input-expected.txt:
* platform/wpe/imported/w3c/web-platform-tests/encrypted-media/clearkey-mp4-unique-origin-expected.txt:
    Since r226621 initDataType "cenc" is supported in clearKey, so we should update the expected result of these tests.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226965 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months ago[EME] some fixes in clearKey-encrypted-cenc-event LayoutTest
commit-queue@webkit.org [Tue, 16 Jan 2018 08:04:12 +0000 (08:04 +0000)]
[EME] some fixes in clearKey-encrypted-cenc-event LayoutTest
https://bugs.webkit.org/show_bug.cgi?id=181526

Patch by Yacine Bandou <yacine.bandou_ext@softathome.com> on 2018-01-16
Reviewed by Xabier Rodriguez-Calvar.

Update the TestExpectation files as it was done with clearKey-message-cenc-event
LayoutTest, the test should be pass in all platforms except gtk and mac.
Use const or let instead of var in JS when it's possible.

* TestExpectations:
* media/encrypted-media/clearKey/clearKey-encrypted-cenc-event.html:
* media/media-source/media-source-loader-simple.js:
* platform/gtk/TestExpectations:
* platform/mac/TestExpectations:
* platform/wpe/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226964 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months agoWeb Inspector: TabBar redesign: add context menu to TabBar for toggling available...
mattbaker@apple.com [Tue, 16 Jan 2018 07:03:57 +0000 (07:03 +0000)]
Web Inspector: TabBar redesign: add context menu to TabBar for toggling available tabs
https://bugs.webkit.org/show_bug.cgi?id=181448
<rdar://problem/36383298>

Reviewed by Devin Rousso.

* UserInterface/Base/Main.js:
(WI.loaded):
(WI.contentLoaded):
Reorder production tab classes and default (open) tabs.
(WI.registerTabClass): Removed.
Not used.

* UserInterface/Base/Object.js:
Remove notification that is no longer used.

* UserInterface/Views/GeneralTabBarItem.js:
(WI.GeneralTabBarItem):
(WI.GeneralTabBarItem.prototype.set title):
(WI.GeneralTabBarItem.prototype._handleContextMenuEvent): Deleted.
Remove per-tab context menu (provided Close Tab and Close Other Tabs).

* UserInterface/Views/NewTabContentView.js:
(WI.NewTabContentView):

* UserInterface/Views/TabBar.js:
(WI.TabBar):
(WI.TabBar.prototype._handleContextMenu):

* UserInterface/Views/TabBrowser.js:
(WI.TabBrowser):
(WI.TabBrowser._handleNewTabContextMenu): Deleted.
No longer needed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226963 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months agoSupport for preconnect Link headers
yoav@yoav.ws [Tue, 16 Jan 2018 06:44:41 +0000 (06:44 +0000)]
Support for preconnect Link headers
https://bugs.webkit.org/show_bug.cgi?id=181657

Reviewed by Darin Adler.

Source/WebCore:

Move the preconnect functionality into its own function, and
also call this function when Link headers are processed.

Test: http/tests/preconnect/link-header-rel-preconnect-http.php

* loader/LinkLoader.cpp:
(WebCore::LinkLoader::loadLinksFromHeader): Call preconnect.
(WebCore::LinkLoader::preconnect): Preconnect to a host functionality moved here.
(WebCore::LinkLoader::preload): Renamed `preloadIfNeeded` to `preload`.
(WebCore::LinkLoader::loadLink): Call preconnect.
* loader/LinkLoader.h:

LayoutTests:

Add test to see Link preconnect headers trigger a connection.

* http/tests/preconnect/link-header-rel-preconnect-http-expected.txt: Added.
* http/tests/preconnect/link-header-rel-preconnect-http.php: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226962 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months agoUnreviewed GTK layout test gardening
mcatanzaro@igalia.com [Tue, 16 Jan 2018 02:44:14 +0000 (02:44 +0000)]
Unreviewed GTK layout test gardening

* TestExpectations:
* platform/gtk/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226961 268f45cc-cd09-0410-ab3c-d52691b4dbfc

20 months agoΩUpdate the default hyperlink for Speedometer to Speedometer 2.0.
rniwa@webkit.org [Tue, 16 Jan 2018 02:36:48 +0000 (02:36 +0000)]
ΩUpdate the default hyperlink for Speedometer to Speedometer 2.0.

Rubber-stamped by Saam Barati.

* index.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226960 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoUnreviewed, add Thibault to contributors.json
mcatanzaro@igalia.com [Tue, 16 Jan 2018 02:20:11 +0000 (02:20 +0000)]
Unreviewed, add Thibault to contributors.json

It's a bit weird to add someone here before their first contribution, but this affects
Bugzilla autocomplete, and that is serious business.

* Scripts/webkitpy/common/config/contributors.json:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226959 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoImprove use of ExportMacros
mcatanzaro@igalia.com [Mon, 15 Jan 2018 23:49:02 +0000 (23:49 +0000)]
Improve use of ExportMacros
https://bugs.webkit.org/show_bug.cgi?id=181652

Reviewed by Konstantin Tokarev.

Source/bmalloc:

Disable BEXPORT on Linux ports.

* bmalloc/BExport.h: Check for BUSE(EXPORT_MACROS).
* bmalloc/BPlatform.h: Add BUSE(EXPORT_MACROS) and define it on macOS and iOS.

Source/JavaScriptCore:

* API/JSBase.h: Update a comment.
* inspector/InspectorBackendDispatcher.h: Use a better, yet equivalent, WTF macro.
* runtime/JSExportMacros.h: Simplify the #defines in this file.

Source/WebCore:

Remove a comment.

* platform/PlatformExportMacros.h:

Source/WebCore/PAL:

Remove a comment.

* pal/ExportMacros.h:

Source/WTF:

* wtf/ExportMacros.h: Simplify the #defines in this file.
* wtf/Platform.h: Remove unneeded define. Remove comment that expects all ports to
  eventually enable the export macros. WPE will never want these. We don't currently want
  them for GTK either, though how we link GTK has been in flux recently.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226958 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoimage-rendering should affect scaling of border-image
annulen@yandex.ru [Mon, 15 Jan 2018 22:56:24 +0000 (22:56 +0000)]
image-rendering should affect scaling of border-image
https://bugs.webkit.org/show_bug.cgi?id=169440

Reviewed by Michael Catanzaro.

Source/WebCore:

Test: fast/borders/border-image-pixelated.html

* rendering/style/NinePieceImage.cpp:
(WebCore::NinePieceImage::paint):

LayoutTests:

* fast/borders/border-image-pixelated-expected.html: Added.
* fast/borders/border-image-pixelated.html: Added.
* fast/borders/resources/border-image-pixelated.png: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226957 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoRemove makePoisonedUnique
jfbastien@apple.com [Mon, 15 Jan 2018 22:02:45 +0000 (22:02 +0000)]
Remove makePoisonedUnique
https://bugs.webkit.org/show_bug.cgi?id=181630
<rdar://problem/36498623>

Reviewed by Mark Lam.

Source/JavaScriptCore:

I added a conversion from std::unique_ptr, so we can just use
std::make_unique and it'll auto-poison when converted.

* bytecode/CodeBlock.h:
(JSC::CodeBlock::makePoisonedUnique): Deleted.
* runtime/JSGlobalObject.cpp:
(JSC::JSGlobalObject::init):
* runtime/JSGlobalObject.h:
(JSC::JSGlobalObject::makePoisonedUnique): Deleted.

Source/WTF:

I added a conversion from std::unique_ptr, so we can just use
std::make_unique and it'll auto-poison when converted.

* wtf/PoisonedUniquePtr.h:
(WTF::makePoisonedUnique): Deleted.

Tools:

Update tests.

* TestWebKitAPI/Tests/WTF/PoisonedUniquePtr.cpp:
(TestWebKitAPI::TEST):
* TestWebKitAPI/Tests/WTF/PoisonedUniquePtrForNonTriviallyDestructibleArrays.cpp:
(TestWebKitAPI::TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226956 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago2.19.3 ACCELERATED_2D_CANVAS support is broken
tpopela@redhat.com [Mon, 15 Jan 2018 21:24:54 +0000 (21:24 +0000)]
2.19.3 ACCELERATED_2D_CANVAS support is broken
https://bugs.webkit.org/show_bug.cgi?id=180799

Reviewed by Michael Catanzaro.

* html/canvas/CanvasRenderingContext2DBase.cpp:
(WebCore::CanvasRenderingContext2DBase::didDraw):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226955 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[Curl] Enable HTTP/2
commit-queue@webkit.org [Mon, 15 Jan 2018 21:21:53 +0000 (21:21 +0000)]
[Curl] Enable HTTP/2
https://bugs.webkit.org/show_bug.cgi?id=181551

Patch by Basuke Suzuki <Basuke.Suzuki@sony.com> on 2018-01-15
Reviewed by Michael Catanzaro.

Start supporting HTTP/2 protocol. The first step is just enabling the HTTP/2 on Curl backend.
Next step will be to enable multiplexing feature.

No new tests because we don't have HTTP/2 test backend yet.

* platform/network/curl/CurlContext.cpp:
(WebCore::CurlContext::isHttp2Enabled const):
(WebCore::CurlHandle::enableHttp):
(WebCore::CurlHandle::enableHttpGetRequest):
(WebCore::CurlHandle::enableHttpHeadRequest):
(WebCore::CurlHandle::enableHttpPostRequest):
(WebCore::CurlHandle::enableHttpPutRequest):
(WebCore::CurlHandle::setHttpCustomRequest):
* platform/network/curl/CurlContext.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226954 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoUse a helper function for checked arithmetic in WebGL validation
dino@apple.com [Mon, 15 Jan 2018 20:43:39 +0000 (20:43 +0000)]
Use a helper function for checked arithmetic in WebGL validation
https://bugs.webkit.org/show_bug.cgi?id=181620
<rdar://problem/36485879>

Reviewed by Eric Carlson.

Eric recommended using a templated helper function to do
a common arithmetic check in WebGL validation.

* html/canvas/WebGL2RenderingContext.cpp:
(WebCore::WebGL2RenderingContext::validateIndexArrayConservative):
* html/canvas/WebGLRenderingContext.cpp:
(WebCore::WebGLRenderingContext::validateIndexArrayConservative):
* html/canvas/WebGLRenderingContextBase.cpp:
(WebCore::WebGLRenderingContextBase::validateIndexArrayPrecise):
(WebCore::WebGLRenderingContextBase::validateDrawArrays):
(WebCore::WebGLRenderingContextBase::validateSimulatedVertexAttrib0):
(WebCore::WebGLRenderingContextBase::simulateVertexAttrib0):
* html/canvas/WebGLRenderingContextBase.h:
(WebCore::WebGLRenderingContextBase::checkedAddAndMultiply): New helper.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226953 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoUse traits for animation timing functions
dino@apple.com [Mon, 15 Jan 2018 19:16:03 +0000 (19:16 +0000)]
Use traits for animation timing functions
https://bugs.webkit.org/show_bug.cgi?id=181651
<rdar://problem/36525328>

Reviewed by Antoine Quint.

Use the type traits for TimingFunction classes, so
we can is<> and downcast<>.

* css/CSSComputedStyleDeclaration.cpp:
(WebCore::createTimingFunctionValue):
* platform/animation/TimingFunction.cpp:
(WebCore::TimingFunction::transformTime const):
* platform/animation/TimingFunction.h:
* platform/graphics/ca/GraphicsLayerCA.cpp:
(WebCore::animationHasStepsTimingFunction):
(WebCore::animationHasFramesTimingFunction):
* platform/graphics/ca/cocoa/PlatformCAAnimationCocoa.mm:
(WebCore::toCAMediaTimingFunction):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226952 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoRealtimeMediaSource should be ThreadSafeRefCounted
commit-queue@webkit.org [Mon, 15 Jan 2018 17:54:00 +0000 (17:54 +0000)]
RealtimeMediaSource should be ThreadSafeRefCounted
https://bugs.webkit.org/show_bug.cgi?id=181649

Patch by Youenn Fablet <youenn@apple.com> on 2018-01-15
Reviewed by Eric Carlson.

Difficult to write a test as this is really racy.
RealtimeIncomingVideoSourceCocoa::OnFrame is taking a reference on a background thread
to send a task to the main thread.
This requires it to be thread safe ref counted.

* platform/mediastream/RealtimeMediaSource.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226951 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[GTK][JHBuild] Tools/gtk/install-dependencies: Package libxfont does not exist anymor...
aperez@igalia.com [Mon, 15 Jan 2018 17:36:15 +0000 (17:36 +0000)]
[GTK][JHBuild] Tools/gtk/install-dependencies: Package libxfont does not exist anymore in Arch Linux
https://bugs.webkit.org/show_bug.cgi?id=181476

Reviewed by Michael Catanzaro.

Update to xorg-server 1.19.6 which depends on libxfont2, and change
the lists of packages to install libxfont2-devel (or equivalent).

* gtk/install-dependencies:
* gtk/jhbuild.modules:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226950 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoUnreviewed GTK gardening
philn@webkit.org [Mon, 15 Jan 2018 15:49:25 +0000 (15:49 +0000)]
Unreviewed GTK gardening

* platform/gtk/TestExpectations: Skip
media/media-fullscreen-pause-inline.html as advised by Eric
Carlson.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226949 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoPrevent useless MediaPlayer mute state notifications
philn@webkit.org [Mon, 15 Jan 2018 15:27:23 +0000 (15:27 +0000)]
Prevent useless MediaPlayer mute state notifications
https://bugs.webkit.org/show_bug.cgi?id=181646

Reviewed by Carlos Garcia Campos.

Source/WebCore:

On GTK port the mute change notification was triggering the test
runner to think the whole page mute state had changed and that
media elements were muted. The simplest solution is to propagate
the notification only if the state actually changed.

* platform/graphics/MediaPlayer.cpp:
(WebCore::MediaPlayer::muteChanged): Propagate mute notification
only if the mute state actually changed.

LayoutTests:

Unflag tests depending on muted state and media session
restrictions, most of those are now passing.

* platform/gtk/TestExpectations:
* platform/gtk/media/video-controls-no-scripting-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226948 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[GStreamer] Don't wait for draw condition variable when shutting down.
commit-queue@webkit.org [Mon, 15 Jan 2018 10:50:44 +0000 (10:50 +0000)]
[GStreamer] Don't wait for draw condition variable when shutting down.
https://bugs.webkit.org/show_bug.cgi?id=180978

Patch by Sebastian Dröge <sebastian@centricular.com> on 2018-01-15
Reviewed by Carlos Garcia Campos.

* platform/graphics/gstreamer/MediaPlayerPrivateGStreamerBase.cpp:
(WebCore::MediaPlayerPrivateGStreamerBase::triggerRepaint):
(WebCore::MediaPlayerPrivateGStreamerBase::cancelRepaint):
* platform/graphics/gstreamer/MediaPlayerPrivateGStreamerBase.h:
By also waiting for the draw condition variable while shutting down,
it is possible that the GStreamer video sink is waiting for the main
thread to actually render the current frame, while at the same time
the main thread is waiting for the GStreamer video sink to shut down,
resulting in a deadlock.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226947 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoASSERTION FAILED: m_ptr under WebKit::CacheStorage::Caches::writeRecord
commit-queue@webkit.org [Mon, 15 Jan 2018 09:36:36 +0000 (09:36 +0000)]
ASSERTION FAILED: m_ptr under WebKit::CacheStorage::Caches::writeRecord
https://bugs.webkit.org/show_bug.cgi?id=181401
<rdar://problem/36379022>

Patch by Youenn Fablet <youenn@apple.com> on 2018-01-15
Reviewed by Darin Adler.

m_isInitialized is true but m_storage is set to nullptr when calling writeRecord.
The only case seems to be if:
- We are doing persistent storage
- We are traversing the storage to do the initialization. At that point m_storage is set to a correct value.
- clearMemoryRepresentation is called, thus setting m_storage to nullptr and m_isInitialized to false.
We fix this by making clearMemoryRepresentation a no-op if we are initializing the caches.
clearMemoryRepresentation is about cleaning the in-memory information of the caches and the memory representation is empty at init time.
Nullifying m_storage is a memory consumption optimization.

* NetworkProcess/cache/CacheStorageEngineCaches.cpp:
(WebKit::CacheStorage::Caches::clearMemoryRepresentation):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226946 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoREGRESSION(r226266): [GTK] RELEASE_ASSERT(reservedZoneSize >= minimumReservedZoneSize...
mcatanzaro@igalia.com [Mon, 15 Jan 2018 08:45:13 +0000 (08:45 +0000)]
REGRESSION(r226266): [GTK] RELEASE_ASSERT(reservedZoneSize >= minimumReservedZoneSize) in JSC::VM::updateStackLimits
https://bugs.webkit.org/show_bug.cgi?id=181438
<rdar://problem/36376724>

Reviewed by Carlos Garcia Campos.

.:

Build JSC as a shared library.

Stop using -fvisibility=hidden. This is a shame, but it is causing problems that I don't
know how to fix.

* Source/cmake/OptionsGTK.cmake:
* Source/cmake/WebKitCompilerFlags.cmake:

Source/JavaScriptCore:

Roll out the functional changes of r226266. We'll keep the minor CMake library type setting
cleanup, but we have to switch back to building JSC only as a shared library, and we have to
get rid of the version script.

* PlatformGTK.cmake:
* javascriptcoregtk-symbols.map: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226945 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[GTK][WPE] Add support for unit test expectations
carlosgc@webkit.org [Mon, 15 Jan 2018 08:17:46 +0000 (08:17 +0000)]
[GTK][WPE] Add support for unit test expectations
https://bugs.webkit.org/show_bug.cgi?id=181589

Reviewed by Michael Catanzaro.

We currently have a way to skip tests by annotating them in the api test runner script. The main problem of this
approach is that we skip tests when they fail in the bots and we never notice if they stop failing, keeping the
tests skipped forever. This is indeed the case of several WebKit2 C API tests. Annotating skipped tests in the
script itself is not a good idea either.

This patch adds a generic TestExpectations class for simple tests based on tests with subtests, like our unit
tests, but also WebDriver tests. It parses a json file with the tests and subtests expectations and provides
convenient methods to query them.

* Scripts/run-gtk-tests:
(GtkTestRunner): Remove all Skipped and Slow tests marked here.
* Scripts/run-wpe-tests:
(WPETestRunner): Ditto.
* Scripts/webkitpy/common/test_expectations.py: Added.
(TestExpectations):
(TestExpectations.__init__):
(TestExpectations._port_name_for_expected):
(TestExpectations._expected_value):
(TestExpectations.skipped_tests):
(TestExpectations.skipped_subtests):
(TestExpectations._expectation_value):
(TestExpectations.is_slow):
(TestExpectations.get_expectation):
* Scripts/webkitpy/common/test_expectations_unittest.py: Added.
(MockTestExpectations):
(MockTestExpectations.__init__):
(MockTestExpectations.is_skip):
(ExpectationsTest):
(assert_exp):
(assert_not_exp):
(assert_bad_exp):
(assert_skip):
(test_basic):
(test_skip):
(test_flaky):
(test_build_type):
* TestWebKitAPI/glib/TestExpectations.json: Added.
* glib/api_test_runner.py:
(TestRunner): Remove SkippedTest implementation.
(TestRunner.__init__): Create a TestExpectations.
(TestRunner._test_cases_to_skip): Use TestExpectations to check skipped tests.
(TestRunner._should_run_test_program): Ditto.
(TestRunner._run_test_glib): Use TestExpectations to check if test suite is slow.
(TestRunner._run_test_glib.parse_line.set_test_result): Register also tests passing.
(TestRunner._run_google_test): Use TestExpectations to check if test cases is slow and register tests passing.
(TestRunner.run_tests): Check if actual result is the expected one and register also unexpected passes.
(TestRunner.run_tests.report): Helper to write report to stdout.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226944 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[macOS] [WK2] Unskip some consistently-passing tests
mmaxfield@apple.com [Sun, 14 Jan 2018 23:02:53 +0000 (23:02 +0000)]
[macOS] [WK2] Unskip some consistently-passing tests
https://bugs.webkit.org/show_bug.cgi?id=181226
<rdar://problem/23023848>

Reviewed by Simon Fraser.

According to the flakiness dashboard, these tests consistently pass now.

* platform/mac-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226943 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoUnreviewed. r226928 broke the CLOOP build. This patch fixes the CLOOP build.
sbarati@apple.com [Sun, 14 Jan 2018 19:43:39 +0000 (19:43 +0000)]
Unreviewed. r226928 broke the CLOOP build. This patch fixes the CLOOP build.

* bytecode/CallLinkStatus.cpp:
(JSC::CallLinkStatus::computeFromLLInt):
(JSC::CallLinkStatus::computeExitSiteData):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226942 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoExtend testing added in r226842
cdumez@apple.com [Sun, 14 Jan 2018 07:16:21 +0000 (07:16 +0000)]
Extend testing added in r226842
https://bugs.webkit.org/show_bug.cgi?id=181634

Reviewed by Darin Adler.

I have verified that the behavior is consistent with Chrome and Firefox.

* fast/dom/Window/window-opener-shadowing-expected.txt:
* fast/dom/Window/window-opener-shadowing.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226941 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoReplace all use of ConstExprPoisoned with Poisoned.
mark.lam@apple.com [Sun, 14 Jan 2018 06:11:55 +0000 (06:11 +0000)]
Replace all use of ConstExprPoisoned with Poisoned.
https://bugs.webkit.org/show_bug.cgi?id=181542
<rdar://problem/36442138>

Reviewed by JF Bastien.

Source/JavaScriptCore:

1. All JSC poisons are now defined in JSCPoison.h.

2. Change all clients to use the new poison values via the POISON() macro.

3. The LLInt code has been updated to handle CodeBlock poison.  Some of this code
   uses the t5 temp register, which is not available on the Windows port.
   Fortunately, we don't currently do poisoning on the Windows port yet.  So,
   it will just work for now.

   When poisoning is enabled for the Windows port, this LLInt code will need a
   Windows specific implementation to workaround its lack of a t5 register.

* API/JSAPIWrapperObject.h:
* API/JSCallbackFunction.h:
* API/JSCallbackObject.h:
* JavaScriptCore.xcodeproj/project.pbxproj:
* Sources.txt:
* assembler/MacroAssemblerCodeRef.h:
(JSC::MacroAssemblerCodePtr::emptyValue):
(JSC::MacroAssemblerCodePtr::deletedValue):
* b3/B3LowerMacros.cpp:
* b3/testb3.cpp:
(JSC::B3::testInterpreter):
* bytecode/CodeBlock.h:
(JSC::CodeBlock::instructions):
(JSC::CodeBlock::instructions const):
(JSC::CodeBlock::makePoisonedUnique):
* dfg/DFGOSRExitCompilerCommon.h:
(JSC::DFG::adjustFrameAndStackInOSRExitCompilerThunk):
* dfg/DFGSpeculativeJIT.cpp:
(JSC::DFG::SpeculativeJIT::compileCheckSubClass):
(JSC::DFG::SpeculativeJIT::emitSwitchIntJump):
* ftl/FTLLowerDFGToB3.cpp:
(JSC::FTL::DFG::LowerDFGToB3::compileCheckSubClass):
* jit/JIT.h:
* jit/ThunkGenerators.cpp:
(JSC::virtualThunkFor):
(JSC::nativeForGenerator):
(JSC::boundThisNoArgsFunctionCallGenerator):
* llint/LowLevelInterpreter.asm:
* llint/LowLevelInterpreter32_64.asm:
* llint/LowLevelInterpreter64.asm:
* parser/UnlinkedSourceCode.h:
* runtime/ArrayPrototype.h:
* runtime/CustomGetterSetter.h:
* runtime/DateInstance.h:
* runtime/InternalFunction.h:
* runtime/JSArrayBuffer.h:
* runtime/JSCPoison.cpp: Copied from Source/JavaScriptCore/runtime/JSCPoisonedPtr.cpp.
(JSC::initializePoison):
* runtime/JSCPoison.h:
(): Deleted.
* runtime/JSCPoisonedPtr.cpp: Removed.
* runtime/JSCPoisonedPtr.h: Removed.
* runtime/JSGlobalObject.h:
(JSC::JSGlobalObject::makePoisonedUnique):
* runtime/JSScriptFetchParameters.h:
* runtime/JSScriptFetcher.h:
* runtime/NativeExecutable.h:
* runtime/StructureTransitionTable.h:
(JSC::StructureTransitionTable::map const):
(JSC::StructureTransitionTable::weakImpl const):
* runtime/WriteBarrier.h:
(JSC::WriteBarrier::poison):
* wasm/js/JSToWasm.cpp:
(JSC::Wasm::createJSToWasmWrapper):
* wasm/js/JSWebAssemblyCodeBlock.cpp:
(JSC::JSWebAssemblyCodeBlock::JSWebAssemblyCodeBlock):
* wasm/js/JSWebAssemblyCodeBlock.h:
* wasm/js/JSWebAssemblyInstance.h:
* wasm/js/JSWebAssemblyMemory.h:
* wasm/js/JSWebAssemblyModule.h:
* wasm/js/JSWebAssemblyTable.h:
* wasm/js/WasmToJS.cpp:
(JSC::Wasm::handleBadI64Use):
(JSC::Wasm::wasmToJS):
* wasm/js/WebAssemblyFunctionBase.h:
* wasm/js/WebAssemblyModuleRecord.h:
* wasm/js/WebAssemblyToJSCallee.h:
* wasm/js/WebAssemblyWrapperFunction.h:

Source/WTF:

1. Removed ConstExprPoisoned and its artifacts.

2. Consolidated Poisoned into PoisonedImpl.  PoisonedImpl is not more.

3. Changed all clients of ConstExprPoisoned to use Poisoned instead.

4. Worked around the GCC and Clang compiler bug that confuses an intptr_t&
   template arg with intptr_t.
   See use of std::enable_if_t<Other::isPoisoned> in Poisoned.h.

5. Removed ENABLE(MIXED_POISON) since we now have a workaround (3) that makes it
   possible to use the mixed poison code.

6. Also fixed broken implementation of comparison operators in Poisoned.

* wtf/Bag.h:
* wtf/DumbPtrTraits.h:
(WTF::DumbPtrTraits::poison):
* wtf/DumbValueTraits.h:
(WTF::DumbValueTraits::poison):
* wtf/Poisoned.h:
(WTF::Poisoned::Poisoned):
(WTF::Poisoned::operator== const):
(WTF::Poisoned::operator!= const):
(WTF::Poisoned::operator< const):
(WTF::Poisoned::operator<= const):
(WTF::Poisoned::operator> const):
(WTF::Poisoned::operator>= const):
(WTF::Poisoned::operator=):
(WTF::Poisoned::swap):
(WTF::swap):
(WTF::PoisonedPtrTraits::poison):
(WTF::PoisonedPtrTraits::swap):
(WTF::PoisonedValueTraits::poison):
(WTF::PoisonedValueTraits::swap):
(WTF::PoisonedImpl::PoisonedImpl): Deleted.
(WTF::PoisonedImpl::assertIsPoisoned const): Deleted.
(WTF::PoisonedImpl::assertIsNotPoisoned const): Deleted.
(WTF::PoisonedImpl::unpoisoned const): Deleted.
(WTF::PoisonedImpl::clear): Deleted.
(WTF::PoisonedImpl::operator* const): Deleted.
(WTF::PoisonedImpl::operator-> const): Deleted.
(WTF::PoisonedImpl::bits const): Deleted.
(WTF::PoisonedImpl::operator! const): Deleted.
(WTF::PoisonedImpl::operator bool const): Deleted.
(WTF::PoisonedImpl::operator== const): Deleted.
(WTF::PoisonedImpl::operator!= const): Deleted.
(WTF::PoisonedImpl::operator< const): Deleted.
(WTF::PoisonedImpl::operator<= const): Deleted.
(WTF::PoisonedImpl::operator> const): Deleted.
(WTF::PoisonedImpl::operator>= const): Deleted.
(WTF::PoisonedImpl::operator=): Deleted.
(WTF::PoisonedImpl::swap): Deleted.
(WTF::PoisonedImpl::exchange): Deleted.
(WTF::PoisonedImpl::poison): Deleted.
(WTF::PoisonedImpl::unpoison): Deleted.
(WTF::constExprPoisonRandom): Deleted.
(WTF::makeConstExprPoison): Deleted.
(WTF::ConstExprPoisonedPtrTraits::exchange): Deleted.
(WTF::ConstExprPoisonedPtrTraits::swap): Deleted.
(WTF::ConstExprPoisonedPtrTraits::unwrap): Deleted.
(WTF::ConstExprPoisonedValueTraits::exchange): Deleted.
(WTF::ConstExprPoisonedValueTraits::swap): Deleted.
(WTF::ConstExprPoisonedValueTraits::unwrap): Deleted.
* wtf/PoisonedUniquePtr.h:
(WTF::PoisonedUniquePtr::PoisonedUniquePtr):
(WTF::PoisonedUniquePtr::operator=):
* wtf/Ref.h:
* wtf/RefCountedArray.h:
(WTF::RefCountedArray::RefCountedArray):
* wtf/RefPtr.h:
* wtf/WTFAssertions.cpp:

Tools:

1. Converted tests to using new uintptr_t& poison type.
2. Added tests for Poisoned comparison operators.

* TestWebKitAPI/CMakeLists.txt:
* TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
* TestWebKitAPI/Tests/WTF/ConstExprPoisoned.cpp: Removed.
* TestWebKitAPI/Tests/WTF/Poisoned.cpp:
(TestWebKitAPI::TEST):
(TestWebKitAPI::initializeTestPoison): Deleted.
* TestWebKitAPI/Tests/WTF/PoisonedRef.cpp:
(TestWebKitAPI::TEST):
(TestWebKitAPI::passWithRef):
* TestWebKitAPI/Tests/WTF/PoisonedRefPtr.cpp:
(TestWebKitAPI::TEST):
(TestWebKitAPI::f1):
* TestWebKitAPI/Tests/WTF/PoisonedUniquePtr.cpp:
(TestWebKitAPI::TEST):
(TestWebKitAPI::poisonedPtrFoo):
* TestWebKitAPI/Tests/WTF/PoisonedUniquePtrForNonTriviallyDestructibleArrays.cpp:
(TestWebKitAPI::TEST):
* TestWebKitAPI/Tests/WTF/PoisonedUniquePtrForTriviallyDestructibleArrays.cpp:
(TestWebKitAPI::TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226940 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoWeb Inspector: Styles Redesign: properties should never be semitransparent or crossed...
nvasilyev@apple.com [Sun, 14 Jan 2018 04:06:15 +0000 (04:06 +0000)]
Web Inspector: Styles Redesign: properties should never be semitransparent or crossed out while editing
https://bugs.webkit.org/show_bug.cgi?id=180793
<rdar://problem/36038813>

Reviewed by Devin Rousso.

* UserInterface/Views/SpreadsheetCSSStyleDeclarationEditor.css:
(.spreadsheet-style-declaration-editor .value.editing):
(.spreadsheet-style-declaration-editor .property:matches(.invalid-name, .other-vendor, .overridden):not(.disabled) .content > *):
(.spreadsheet-style-declaration-editor .property.invalid-name:not(.disabled) .content > *):
(.spreadsheet-style-declaration-editor .property.invalid-value:not(.disabled) .content .value):
(.spreadsheet-style-declaration-editor .property.not-inherited .content > *):
* UserInterface/Views/SpreadsheetStyleProperty.js:
(WI.SpreadsheetStyleProperty.prototype._update):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226939 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agocheck-webkit-style: false positive for [whitespace/braces] with os_log format string
ddkilzer@apple.com [Sat, 13 Jan 2018 20:38:04 +0000 (20:38 +0000)]
check-webkit-style: false positive for [whitespace/braces] with os_log format string
<https://webkit.org/b/181606>

Reviewed by Darin Adler.

* Scripts/webkitpy/style/checkers/cpp.py:
(check_spacing): Modify regular expression to ignore
os_log format strings like '%{public}'.
* Scripts/webkitpy/style/checkers/cpp_unittest.py:
(CppStyleTest.test_spacing_before_braces): Add os_log
format string test.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226938 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[JSC] NumberPrototype::extractRadixFromArgs incorrectly cast double to int32_t
ticaiolima@gmail.com [Sat, 13 Jan 2018 15:16:23 +0000 (15:16 +0000)]
[JSC] NumberPrototype::extractRadixFromArgs incorrectly cast double to int32_t
https://bugs.webkit.org/show_bug.cgi?id=181182

Reviewed by Darin Adler.

JSTests:

* bigIntTests.yaml:
* stress/big-int-constructor.js:
* stress/big-int-prototype-to-string-cast-overflow.js: Added.
(assert):
(assertThrowRangeError):
* stress/number-prototype-to-string-cast-overflow.js: Added.
(assert):
(assertThrowRangeError):

Source/JavaScriptCore:

Casting double to integer is undefined behavior when the truncation
results into a value that doesn't fit into integer size, according C++
spec[1]. Thus, we are changing bigIntProtoFuncToString and
numberProtoFuncToString to remove these source of undefined behavior.

[1] - http://en.cppreference.com/w/cpp/language/implicit_conversion

* runtime/BigIntPrototype.cpp:
(JSC::bigIntProtoFuncToString):
* runtime/NumberPrototype.cpp:
(JSC::numberProtoFuncToString):
(JSC::extractRadixFromArgs): Deleted.
(JSC::extractToStringRadixArgument): Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226937 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoMathML Lengths should take zoom level into account
commit-queue@webkit.org [Sat, 13 Jan 2018 12:07:52 +0000 (12:07 +0000)]
MathML Lengths should take zoom level into account
https://bugs.webkit.org/show_bug.cgi?id=180029

Patch by Minsheng Liu <lambda@liu.ms> on 2018-01-13
Reviewed by Frédéric Wang.

Source/WebCore:

The patch applies the effective zoom factor to physical units in toUserUnits().

Test: mathml/presentation/mspace-units-with-zoom.html

* rendering/mathml/RenderMathMLBlock.cpp:
(WebCore::toUserUnits):

LayoutTests:

Add a testcase for MathML lengths where zoom factor is set.

* mathml/presentation/mspace-units-with-zoom-expected.html: Added.
* mathml/presentation/mspace-units-with-zoom.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226936 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[macOS] API tests ChangeAttachmentDataUpdatesWithInPlaceDisplay and InPlaceVideoAttac...
wenson_hsieh@apple.com [Sat, 13 Jan 2018 05:30:57 +0000 (05:30 +0000)]
[macOS] API tests ChangeAttachmentDataUpdatesWithInPlaceDisplay and InPlaceVideoAttachmentInsertionWithinList are failing
https://bugs.webkit.org/show_bug.cgi?id=181626

Reviewed by Joseph Pecoraro.

Make the web view for WKAttachmentTests slightly bigger to avoid size clamping behavior introduced in r226915.
Additionally, write a new API test covering the change in r226915.

* TestWebKitAPI/Tests/WebKitCocoa/WKAttachmentTests.mm:
(webViewForTestingAttachments):
(TestWebKitAPI::TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226935 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoASSERTION FAILED: registration || isTerminating() in WebCore::SWServerWorker::skipWai...
cdumez@apple.com [Sat, 13 Jan 2018 04:09:43 +0000 (04:09 +0000)]
ASSERTION FAILED: registration || isTerminating() in WebCore::SWServerWorker::skipWaiting()
https://bugs.webkit.org/show_bug.cgi?id=181603
<rdar://problem/36476050>

Reviewed by Youenn Fablet.

No new tests, covered by existing tests that crash flakily.

* workers/service/server/SWServer.cpp:
(WebCore::SWServer::terminateWorkerInternal):
If the connection to the context process is gone, make sure we make the worker as terminated
so that it does not stay in Running state and in SWServer::m_runningOrTerminatingWorkers.

* workers/service/server/SWServerRegistration.cpp:
(WebCore::SWServerRegistration::~SWServerRegistration):
Add assertions to make sure none of the registration's workers are still running when
the registration is destroyed.

(WebCore::SWServerRegistration::updateRegistrationState):
Make sure registration workers that are overwritten are not still running.

* workers/service/server/SWServerWorker.cpp:
(WebCore::SWServerWorker::setState):
If a worker's state is set to redundant, make sure we also terminate it.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226934 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoUnreviewed, speculative build fix after r226899.
wenson_hsieh@apple.com [Sat, 13 Jan 2018 01:11:34 +0000 (01:11 +0000)]
Unreviewed, speculative build fix after r226899.

Add an empty implementation for PageClient::startDrag.

* UIProcess/PageClient.h:
(WebKit::PageClient::startDrag):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226933 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoTouch events not received with initial-scale=1 viewport
simon.fraser@apple.com [Sat, 13 Jan 2018 01:03:42 +0000 (01:03 +0000)]
Touch events not received with initial-scale=1 viewport
https://bugs.webkit.org/show_bug.cgi?id=181566
rdar://problem/36460221

Reviewed by Tim Horton.

Add a testcase for this touch event case.

* fast/events/touch/ios/target-taller-than-view-expected.txt: Added.
* fast/events/touch/ios/target-taller-than-view.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226932 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoUnreviewed, rolling out r226927.
commit-queue@webkit.org [Sat, 13 Jan 2018 00:46:06 +0000 (00:46 +0000)]
Unreviewed, rolling out r226927.
https://bugs.webkit.org/show_bug.cgi?id=181621

Breaks 32-bit and iOS release for some reason that i don't
understand yet (Requested by dino on #webkit).

Reverted changeset:

"Use a helper function for checked arithmetic in WebGL
validation"
https://bugs.webkit.org/show_bug.cgi?id=181620
https://trac.webkit.org/changeset/226927

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226931 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoData URL fonts split in the middle of an alphabet cause random letters to disappear
mmaxfield@apple.com [Sat, 13 Jan 2018 00:45:26 +0000 (00:45 +0000)]
Data URL fonts split in the middle of an alphabet cause random letters to disappear
https://bugs.webkit.org/show_bug.cgi?id=175845
<rdar://problem/33996578>

Reviewed by Brent Fulgham.

Source/WebCore:

It is fairly common practice for a font foundry to split a font up into two files such that a semi-random
half of the alphabet is present in one of the files, and the other half is present in the other file. This
practice involves representing the files as data URLs, so as to minimize the time it takes to load them.

Because resource loading is asynchronous (even for data URLs), it is possible today to get a paint after
the first file is loaded but before the second file is loaded. Indeed, because of the way font fallback
works, we will never start loading the second file until a layout has occurred with the first font.

Because a site usually only uses this pattern for a handful of fonts, and I've never seen this pattern
being used for CJK fonts, it isn't very expensive to opportunistically decode these data URLs eagerly.
Using this method doesn't actually guarantee that the two fonts will load in between successive paints,
but it at least makes this much more likely. This patch implements this strategy, along with a size
threshold to make sure that we won't decode any super large data URLs when it isn't necessary.

Test: fast/text/font-load-data-partitioned-alphabet.html

* css/CSSFontFace.cpp:
(WebCore::CSSFontFace::opportunisticallyStartFontDataURLLoading):
* css/CSSFontFace.h:
* css/CSSFontFaceSource.cpp:
(WebCore::CSSFontFaceSource::opportunisticallyStartFontDataURLLoading):
* css/CSSFontFaceSource.h:
* css/CSSFontSelector.cpp:
(WebCore::CSSFontSelector::opportunisticallyStartFontDataURLLoading):
* css/CSSFontSelector.h:
* platform/graphics/FontCascadeFonts.cpp:
(WebCore::opportunisticallyStartFontDataURLLoading):
(WebCore::FontCascadeFonts::glyphDataForVariant):
* platform/graphics/FontSelector.h:

LayoutTests:

Make sure that the requests for both fonts are sent before either of the responses are received.

* fast/text/font-load-data-partitioned-alphabet-expected.txt: Added.
* fast/text/font-load-data-partitioned-alphabet.html: Added.
* platform/mac-wk1/fast/text/font-load-data-partitioned-alphabet-expected.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226930 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoHistory state should be updated during client redirects with asynchronous policy...
achristensen@apple.com [Sat, 13 Jan 2018 00:39:34 +0000 (00:39 +0000)]
History state should be updated during client redirects with asynchronous policy decisions
https://bugs.webkit.org/show_bug.cgi?id=181358
<rdar://problem/35547689>

Reviewed by Andy Estes.

Source/WebCore:

When decidePolicyForNavigationAction is responded to asynchronously during a client redirect,
HistoryController::updateForRedirectWithLockedBackForwardList does not update the history because
the document loader has not been marked as a client redirect because the FrameLoader only looks
at its provisional document loader to mark it as a client redirect.  When decidePolicyForNavigationAction
is responded to asynchronously, though, the FrameLoader's provisional document loader has moved to
its policy document loader.  To get both asynchronous and synchronous cases, let's just mark the document
loader as a client redirect whether it's the provisional or policy document loader.

Covered by a new API test.

* loader/FrameLoader.cpp:
(WebCore::FrameLoader::loadURL):
(WebCore::FrameLoader::loadPostRequest):

Tools:

* TestWebKitAPI/Tests/WebKit/WKBackForwardList.mm:
(-[AsyncPolicyDecisionDelegate webView:didFinishNavigation:]):
(-[AsyncPolicyDecisionDelegate webView:decidePolicyForNavigationAction:decisionHandler:]):
(TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226929 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoMove ExitProfile to UnlinkedCodeBlock so it can be shared amongst CodeBlocks backed...
sbarati@apple.com [Sat, 13 Jan 2018 00:36:37 +0000 (00:36 +0000)]
Move ExitProfile to UnlinkedCodeBlock so it can be shared amongst CodeBlocks backed by the same UnlinkedCodeBlock
https://bugs.webkit.org/show_bug.cgi?id=181545

Reviewed by Michael Saboff.

This patch follows the theme of putting optimization profiling information on
UnlinkedCodeBlock. This allows the unlinked code cache to remember OSR exit data.
This often leads to the first compile of a CodeBlock, backed by an UnlinkedCodeBlock
pulled from the code cache, making better compilation decisions, usually
resulting in fewer exits, and fewer recompilations.

This is a 1% Speedometer progression in my testing.

* bytecode/BytecodeDumper.cpp:
(JSC::BytecodeDumper<CodeBlock>::dumpProfilesForBytecodeOffset):
* bytecode/CallLinkStatus.cpp:
(JSC::CallLinkStatus::computeFromLLInt):
(JSC::CallLinkStatus::computeFor):
(JSC::CallLinkStatus::computeExitSiteData):
(JSC::CallLinkStatus::computeDFGStatuses):
* bytecode/CallLinkStatus.h:
* bytecode/CodeBlock.h:
(JSC::CodeBlock::addFrequentExitSite): Deleted.
(JSC::CodeBlock::hasExitSite const): Deleted.
(JSC::CodeBlock::exitProfile): Deleted.
* bytecode/DFGExitProfile.cpp:
(JSC::DFG::ExitProfile::add):
(JSC::DFG::QueryableExitProfile::initialize):
* bytecode/DFGExitProfile.h:
(JSC::DFG::ExitProfile::hasExitSite const):
* bytecode/GetByIdStatus.cpp:
(JSC::GetByIdStatus::hasExitSite):
(JSC::GetByIdStatus::computeFor):
(JSC::GetByIdStatus::computeForStubInfo):
* bytecode/GetByIdStatus.h:
* bytecode/PutByIdStatus.cpp:
(JSC::PutByIdStatus::hasExitSite):
(JSC::PutByIdStatus::computeFor):
(JSC::PutByIdStatus::computeForStubInfo):
* bytecode/PutByIdStatus.h:
* bytecode/UnlinkedCodeBlock.cpp:
(JSC::UnlinkedCodeBlock::livenessAnalysisSlow):
* bytecode/UnlinkedCodeBlock.h:
(JSC::UnlinkedCodeBlock::hasExitSite const):
(JSC::UnlinkedCodeBlock::hasExitSite):
(JSC::UnlinkedCodeBlock::exitProfile):
* dfg/DFGByteCodeParser.cpp:
(JSC::DFG::ByteCodeParser::InlineStackEntry::InlineStackEntry):
* dfg/DFGGraph.h:
(JSC::DFG::Graph::hasGlobalExitSite):
(JSC::DFG::Graph::hasExitSite):
* dfg/DFGLICMPhase.cpp:
(JSC::DFG::LICMPhase::attemptHoist):
* dfg/DFGOSRExitBase.cpp:
(JSC::DFG::OSRExitBase::considerAddingAsFrequentExitSiteSlow):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226928 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoUse a helper function for checked arithmetic in WebGL validation
dino@apple.com [Sat, 13 Jan 2018 00:30:02 +0000 (00:30 +0000)]
Use a helper function for checked arithmetic in WebGL validation
https://bugs.webkit.org/show_bug.cgi?id=181620
<rdar://problem/36485879>

Reviewed by Eric Carlson.

Eric recommended using a templated helper function to do
a common arithmetic check in WebGL validation.

* html/canvas/WebGL2RenderingContext.cpp:
(WebCore::WebGL2RenderingContext::validateIndexArrayConservative):
* html/canvas/WebGLRenderingContext.cpp:
(WebCore::WebGLRenderingContext::validateIndexArrayConservative):
* html/canvas/WebGLRenderingContextBase.cpp:
(WebCore::WebGLRenderingContextBase::checkedAddAndMultiply): New helper.
(WebCore::WebGLRenderingContextBase::validateIndexArrayPrecise):
(WebCore::WebGLRenderingContextBase::validateDrawArrays):
(WebCore::WebGLRenderingContextBase::validateSimulatedVertexAttrib0):
(WebCore::WebGLRenderingContextBase::simulateVertexAttrib0):
* html/canvas/WebGLRenderingContextBase.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226927 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[iOS] Remove unneeded accessibility-related sandbox rules
bfulgham@apple.com [Sat, 13 Jan 2018 00:28:11 +0000 (00:28 +0000)]
[iOS] Remove unneeded accessibility-related sandbox rules
https://bugs.webkit.org/show_bug.cgi?id=181619
<rdar://problem/36485356>

Reviewed by Eric Carlson.

Remove a number of sandbox exceptions that were in place for accessibility support. These are
not needed in the WebContent process, since Safari (not WebKit) handles the accessibility
interactions.

* Resources/SandboxProfiles/ios/com.apple.WebKit.WebContent.sb:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226926 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoPoisonedWriteBarrier
jfbastien@apple.com [Fri, 12 Jan 2018 23:47:58 +0000 (23:47 +0000)]
PoisonedWriteBarrier
https://bugs.webkit.org/show_bug.cgi?id=181599
<rdar://problem/36474351>

Reviewed by Mark Lam.

Source/JavaScriptCore:

Allow poisoning of WriteBarrier objects, and use this for
WebAssembly because it is perf-neutral, at least on WasmBench on
my MBP. If it indeed is perf-neutral according to the bots, start
using it in more performance-sensitive places.

* heap/HandleTypes.h:
* heap/SlotVisitor.h:
* heap/SlotVisitorInlines.h:
(JSC::SlotVisitor::append):
(JSC::SlotVisitor::appendHidden):
* runtime/JSCJSValue.h:
* runtime/JSCPoison.h:
* runtime/Structure.h:
* runtime/StructureInlines.h:
(JSC::Structure::setPrototypeWithoutTransition):
(JSC::Structure::setGlobalObject):
(JSC::Structure::setPreviousID):
* runtime/WriteBarrier.h:
(JSC::WriteBarrierBase::copyFrom):
(JSC::WriteBarrierBase::get const):
(JSC::WriteBarrierBase::operator* const):
(JSC::WriteBarrierBase::operator-> const):
(JSC::WriteBarrierBase::clear):
(JSC::WriteBarrierBase::slot):
(JSC::WriteBarrierBase::operator bool const):
(JSC::WriteBarrierBase::setWithoutWriteBarrier):
(JSC::WriteBarrierBase::unvalidatedGet const):
(JSC::operator==):
* runtime/WriteBarrierInlines.h:
(JSC::Traits>::set):
(JSC::Traits>::setMayBeNull):
(JSC::Traits>::setEarlyValue):
(JSC::DumbValueTraits<Unknown>>::set):
* wasm/WasmInstance.h:
* wasm/js/JSWebAssemblyInstance.cpp:
(JSC::JSWebAssemblyInstance::JSWebAssemblyInstance):
(JSC::JSWebAssemblyInstance::finishCreation):
(JSC::JSWebAssemblyInstance::visitChildren):
(JSC::JSWebAssemblyInstance::create):
* wasm/js/JSWebAssemblyInstance.h:
(JSC::JSWebAssemblyInstance::offsetOfPoisonedCallee):
* wasm/js/JSWebAssemblyMemory.h:
* wasm/js/JSWebAssemblyModule.h:
* wasm/js/JSWebAssemblyTable.cpp:
(JSC::JSWebAssemblyTable::JSWebAssemblyTable):
(JSC::JSWebAssemblyTable::grow):
(JSC::JSWebAssemblyTable::clearFunction):
* wasm/js/JSWebAssemblyTable.h:
* wasm/js/WasmToJS.cpp:
(JSC::Wasm::materializeImportJSCell):
(JSC::Wasm::handleBadI64Use):
(JSC::Wasm::wasmToJS):
* wasm/js/WebAssemblyFunctionBase.h:
* wasm/js/WebAssemblyModuleRecord.cpp:
(JSC::WebAssemblyModuleRecord::link):
(JSC::WebAssemblyModuleRecord::evaluate):
* wasm/js/WebAssemblyModuleRecord.h:
* wasm/js/WebAssemblyToJSCallee.h:
* wasm/js/WebAssemblyWrapperFunction.h:

Source/WTF:

Supporting changes needed to allow poisoning of WriteBarrier
objects.

* WTF.xcodeproj/project.pbxproj:
* wtf/DumbPtrTraits.h:
* wtf/DumbValueTraits.h: Copied from Source/WTF/wtf/DumbPtrTraits.h.
(WTF::DumbValueTraits::exchange):
(WTF::DumbValueTraits::swap):
(WTF::DumbValueTraits::unwrap):
* wtf/Forward.h:
* wtf/Poisoned.h:
(WTF::ConstExprPoisonedValueTraits::exchange):
(WTF::ConstExprPoisonedValueTraits::swap):
(WTF::ConstExprPoisonedValueTraits::unwrap):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226920 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[Cocoa] CTFontCopyDefaultCascadeListForLanguages() can return nullptr
mmaxfield@apple.com [Fri, 12 Jan 2018 23:35:55 +0000 (23:35 +0000)]
[Cocoa] CTFontCopyDefaultCascadeListForLanguages() can return nullptr
https://bugs.webkit.org/show_bug.cgi?id=181615
<rdar://problem/36334637>

Reviewed by Jon Lee.

Speculative fix. We are getting crash reports saying that this call can return nullptr, and we
don't check for it.

No new tests because I couldn't find the specific input that causes it to return nullptr. (I
tried running this code with every 0, 1, and 2 length locale string, every weight value, and
every italic value, and couldn't get it to crash. I also inspected the code to figure out what
values would cause it to return nullptr, and I couldn't find anything other than if the system
has a totally busted font setup.)

* platform/graphics/cocoa/FontDescriptionCocoa.cpp:
(WebCore::SystemFontDatabase::computeCascadeList):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226919 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agowebkitpy: Refactor simulator code (Part 4)
jbedard@apple.com [Fri, 12 Jan 2018 23:29:04 +0000 (23:29 +0000)]
webkitpy: Refactor simulator code (Part 4)
https://bugs.webkit.org/show_bug.cgi?id=180555
<rdar://problem/36131381>

Reviewed by Aakash Jain.

Move new_simulated_device.py to simulated_device.py.

* Scripts/webkitpy/port/ios_simulator.py: Update import statement.
* Scripts/webkitpy/xcode/new_simulated_device.py: Removed.
* Scripts/webkitpy/xcode/new_simulated_device_unittest.py: Removed.
* Scripts/webkitpy/xcode/simulated_device.py: Copied from Tools/Scripts/webkitpy/xcode/new_simulated_device.py.
* Scripts/webkitpy/xcode/simulated_device_unittest.py: Copied from Tools/Scripts/webkitpy/xcode/new_simulated_device_unittest.py.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226918 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[WebGL] Simulated vertexAttrib0 can sometimes cause OUT_OF_MEMORY errors
dino@apple.com [Fri, 12 Jan 2018 23:01:32 +0000 (23:01 +0000)]
[WebGL] Simulated vertexAttrib0 can sometimes cause OUT_OF_MEMORY errors
https://bugs.webkit.org/show_bug.cgi?id=181558
<rdar://problem/36189833>

Reviewed by Eric Carlson.

Source/WebCore:

Very large element indices in the ELEMENT_ARRAY_BUFFER meant that
our simulated vertexAttrib0 buffer might be too large. We need
to check for out-of-memory, but we can also detect some of the issues
earlier in our validation code. Additionally, make sure that we don't
accidentally cast an unsigned to a signed.

Test: fast/canvas/webgl/simulated-vertexAttrib0-invalid-indicies.html

* html/canvas/WebGL2RenderingContext.cpp:
(WebCore::WebGL2RenderingContext::validateIndexArrayConservative): Update validation
code to look for overflow, rather than relying on looking for sign changes.
* html/canvas/WebGLRenderingContext.cpp:
(WebCore::WebGLRenderingContext::validateIndexArrayConservative): Ditto.
* html/canvas/WebGLRenderingContextBase.cpp:
(WebCore::WebGLRenderingContextBase::validateIndexArrayPrecise):
(WebCore::WebGLRenderingContextBase::drawArrays): Check that we were able to simulate.
(WebCore::WebGLRenderingContextBase::drawElements):
(WebCore::WebGLRenderingContextBase::validateSimulatedVertexAttrib0): Update validation code, and
use GC3Duint, since that's what the indicies are.
(WebCore::WebGLRenderingContextBase::simulateVertexAttrib0): Ditto.
(WebCore::WebGLRenderingContextBase::drawArraysInstanced): Check that we were able to simulate.
(WebCore::WebGLRenderingContextBase::drawElementsInstanced):
* html/canvas/WebGLRenderingContextBase.h:

LayoutTests:

* fast/canvas/webgl/simulated-vertexAttrib0-invalid-indicies-expected.txt: Added.
* fast/canvas/webgl/simulated-vertexAttrib0-invalid-indicies.html: Added.
* platform/mac/TestExpectations: Test crashes on Sierra and earlier.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226916 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoLarge in-place attachment elements cause the document width to expand when inserted
wenson_hsieh@apple.com [Fri, 12 Jan 2018 22:00:52 +0000 (22:00 +0000)]
Large in-place attachment elements cause the document width to expand when inserted
https://bugs.webkit.org/show_bug.cgi?id=181614

Reviewed by Dan Bernstein.

Make in-place images and videos have a max-width of 100%, so that large attachments aren't inserted with full
display size, causing the document and viewport width to expand.

* html/HTMLAttachmentElement.cpp:
(WebCore::HTMLAttachmentElement::populateShadowRootIfNecessary):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226915 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoWeb Inspector: Drop support for iOS 7 targets
commit-queue@webkit.org [Fri, 12 Jan 2018 21:47:07 +0000 (21:47 +0000)]
Web Inspector: Drop support for iOS 7 targets
https://bugs.webkit.org/show_bug.cgi?id=181549
<rdar://problem/36444813>

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2018-01-12
Reviewed by Brian Burg.

* Versions/Inspector-iOS-7.0.json: Removed.
* UserInterface/Protocol/Legacy/7.0/InspectorBackendCommands.js: Removed.
Remove protocol snapshot and generated commands for iOS 7.

* UserInterface/Base/Main.js:
* UserInterface/Controllers/BreakpointPopoverController.js:
(WI.BreakpointPopoverController.prototype._createPopoverContent):
* UserInterface/Controllers/DebuggerManager.js:
(WI.DebuggerManager.prototype._setBreakpoint):
* UserInterface/Controllers/SourceMapManager.js:
(WI.SourceMapManager.prototype._loadAndParseSourceMap):
* UserInterface/Models/Instrument.js:
(WI.Instrument.startLegacyTimelineAgent):
* UserInterface/Models/ResourceTimingData.js:
* UserInterface/Models/ScriptTimelineRecord.js:
(WI.ScriptTimelineRecord.EventType.displayName):
* UserInterface/Models/SourceMapResource.js:
(WI.SourceMapResource.prototype.requestContentFromBackend):
* UserInterface/Protocol/CSSObserver.js:
(WI.CSSObserver.prototype.regionLayoutUpdated): Deleted.
* UserInterface/Protocol/RemoteObject.js:
(WI.RemoteObject.fromPayload):
Remove code that was only necessary to support iOS 7.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226914 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoLogged JSON should escape "'s and \'s in strings.
krollin@apple.com [Fri, 12 Jan 2018 21:39:02 +0000 (21:39 +0000)]
Logged JSON should escape "'s and \'s in strings.
https://bugs.webkit.org/show_bug.cgi?id=181608

Reviewed by Brent Fulgham.

Source/WebCore:

No new tests -- no new functionality, just changed logging. The
efficacy of the logging was verified by inspecting its output.

* loader/ResourceLoadObserver.cpp:
(WebCore::ResourceLoadObserver::logUserInteractionWithReducedTimeResolution):

Source/WebKit:

* NetworkProcess/NetworkResourceLoader.cpp:
(WebKit::NetworkResourceLoader::logCookieInformation const):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226913 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoUnreviewed, fixing error in UIKitSPI.h.
megan_gardner@apple.com [Fri, 12 Jan 2018 21:38:47 +0000 (21:38 +0000)]
Unreviewed, fixing error in UIKitSPI.h.

* Platform/spi/ios/UIKitSPI.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226912 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoImplement MultiDocument protocol for restoring focus to a WKWebView
megan_gardner@apple.com [Fri, 12 Jan 2018 21:32:17 +0000 (21:32 +0000)]
Implement MultiDocument protocol for restoring focus to a WKWebView
https://bugs.webkit.org/show_bug.cgi?id=181510

Reviewed by Dan Bernstein.

Support the UIKit protocol for restoring focus to a what previously had focus.
WebKit already knows what node was previously being focused by the DOM, we merely
need to be asked to turn the focus on again.
Resubmitting https://trac.webkit.org/changeset/226826 as it broke internal builds

* Platform/spi/ios/UIKitSPI.h:
* UIProcess/ios/WKContentViewInteraction.h:
* UIProcess/ios/WKContentViewInteraction.mm:
(-[WKContentView _restoreFocusWithToken:]):
(-[WKContentView _preserveFocusWithToken:destructively:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226911 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoWeb Inspector: Support JSX (React) syntax highlighting
commit-queue@webkit.org [Fri, 12 Jan 2018 21:03:22 +0000 (21:03 +0000)]
Web Inspector: Support JSX (React) syntax highlighting
https://bugs.webkit.org/show_bug.cgi?id=181607
<rdar://problem/36442564>

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2018-01-12
Reviewed by Brian Burg.

* UserInterface/Base/MIMETypeUtilities.js:
(WI.mimeTypeForFileExtension):
(WI.fileExtensionForMIMEType):
* UserInterface/Models/Resource.js:
Support the jsx extension and mime types.

* UserInterface/Main.html:
* Scripts/update-codemirror-resources.rb:
* UserInterface/External/CodeMirror/jsx.js: Added.
Include new mode from CodeMirror@d8926768.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226909 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agodrawElements should be invalid if vertexAttrib0 doesn't have data
dino@apple.com [Fri, 12 Jan 2018 21:01:02 +0000 (21:01 +0000)]
drawElements should be invalid if vertexAttrib0 doesn't have data
https://bugs.webkit.org/show_bug.cgi?id=181609
<rdar://problem/36392883>

Reviewed by Antoine Quint.

Source/WebCore:

If a vertex attribute has been enabled, but no data provided, then
draw validation should fail.

Test: fast/canvas/webgl/drawElements-empty-vertex-data.html

* html/canvas/WebGLRenderingContextBase.cpp:
(WebCore::WebGLRenderingContextBase::validateVertexAttributes): If there were
never any data in the vertex buffer, then we incorrectly compared with 0.

LayoutTests:

* fast/canvas/webgl/drawElements-empty-vertex-data-expected.txt: Added.
* fast/canvas/webgl/drawElements-empty-vertex-data.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226908 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoCheckStructure can be incorrectly subsumed by CheckStructureOrEmpty
sbarati@apple.com [Fri, 12 Jan 2018 20:47:44 +0000 (20:47 +0000)]
CheckStructure can be incorrectly subsumed by CheckStructureOrEmpty
https://bugs.webkit.org/show_bug.cgi?id=181177
<rdar://problem/36205704>

Reviewed by Yusuke Suzuki.

JSTests:

* stress/check-structure-ir-ensures-empty-does-not-flow-through.js: Added.
(runNearStackLimit.t):
(runNearStackLimit):
(test.f):
(test):

Source/JavaScriptCore:

The semantics of CheckStructure are such that it does not allow the empty value to flow through it.
However, we may eliminate a CheckStructure if it's preceded by a CheckStructureOrEmpty. This doesn't
have semantic consequences when validation is turned off. However, with validation on, this trips up
our OSR exit machinery that says when an exit is allowed to happen.

Consider the following IR:

a: GetClosureVar // Or any other node that produces BytecodeTop
...
c: CheckStructure(Cell:@a, {s2})
d: PutByOffset(KnownCell:@a, KnownCell:@a, @value)

In the TypeCheckHoistingPhase, we may insert CheckStructureOrEmptys like this:
a: GetClosureVar
e: CheckStructureOrEmpty(@a, {s1})
...
f: CheckStructureOrEmpty(@a, {s2})
c: CheckStructure(Cell:@a, {s2})
d: PutByOffset(KnownCell:@a, KnownCell:@a, @value)

This will cause constant folding to change the IR to:
a: GetClosureVar
e: CheckStructureOrEmpty(@a, {s1})
...
f: CheckStructureOrEmpty(@a, {s2})
d: PutByOffset(KnownCell:@a, KnownCell:@a, @value)

Our mayExit analysis determines that the PutByOffset should not exit. Note
that AI will determine the only value the PutByOffset can see in @a is
the empty value. Because KnownCell filters SpecCell and not SpecCellCheck,
when lowering the PutByOffset, we reach a contradiction in AI and emit
an OSR exit. However, because mayExit said we couldn't exit, we assert.

Note that if we did not run the TypeCheckHoistingPhase on this IR, AI
would have determined we would OSR exit at the second CheckStructure.

This patch makes it so constant folding produces the following IR:
a: GetClosureVar
e: CheckStructureOrEmpty(@a, {s1})
g: AssertNotEmpty(@a)
...
f: CheckStructureOrEmpty(@a, {s2})
h: AssertNotEmpty(@a)
d: PutByOffset(KnownCell:@a, KnownCell:@a, @value)

This modification will cause AI to know we will OSR exit before even reaching
the PutByOffset. Note that in the original IR, the GetClosureVar won't
actually produce the TDZ value. If it did, bytecode would have caused us
to emit a CheckNotEmpty before the CheckStructure/PutByOffset combo. That's
why this bug is about IR bookkeeping and not an actual error in IR analysis.
This patch introduces AssertNotEmpty instead of using CheckNotEmpty to be
more congruous with CheckStructure's semantics of crashing on the empty value
as input (on 64 bit platforms).

* dfg/DFGAbstractInterpreterInlines.h:
(JSC::DFG::AbstractInterpreter<AbstractStateType>::executeEffects):
* dfg/DFGClobberize.h:
(JSC::DFG::clobberize):
* dfg/DFGConstantFoldingPhase.cpp:
(JSC::DFG::ConstantFoldingPhase::foldConstants):
* dfg/DFGDoesGC.cpp:
(JSC::DFG::doesGC):
* dfg/DFGFixupPhase.cpp:
(JSC::DFG::FixupPhase::fixupNode):
* dfg/DFGNodeType.h:
* dfg/DFGPredictionPropagationPhase.cpp:
* 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/FTLLowerDFGToB3.cpp:
(JSC::FTL::DFG::LowerDFGToB3::compileNode):
(JSC::FTL::DFG::LowerDFGToB3::compileAssertNotEmpty):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226907 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoFormDataElement::lengthInBytes should use ThreadableBlobRegistry
commit-queue@webkit.org [Fri, 12 Jan 2018 20:41:55 +0000 (20:41 +0000)]
FormDataElement::lengthInBytes should use ThreadableBlobRegistry
https://bugs.webkit.org/show_bug.cgi?id=181554

Patch by Youenn Fablet <youenn@apple.com> on 2018-01-12
Reviewed by Chris Dumez.

Source/WebCore:

Covered by updated test.

Fix blobRegistry() use and add support for creating a response that may trigger the issue.

* platform/network/FormData.cpp:
(WebCore::FormDataElement::lengthInBytes const): Was using directly blobRegistry() while ThreadableBlobRegistry is more appropriate
in case this is called from workers.
* Modules/fetch/FetchBody.h:
* Modules/fetch/FetchResponse.h:
* fileapi/Blob.h:
* testing/ServiceWorkerInternals.cpp:
(WebCore::ServiceWorkerInternals::createOpaqueWithBlobBodyResponse):
* testing/ServiceWorkerInternals.h:
* testing/ServiceWorkerInternals.idl:

LayoutTests:

* http/tests/workers/service/resources/service-worker-cache-api-worker.js:
(async):
(event.event.request.url.indexOf): Deleted.
(event.event.request.url.endsWith): Deleted.
(event.event.respondWith.promise.then): Deleted.
* http/tests/workers/service/service-worker-cache-api.https-expected.txt:
* http/tests/workers/service/service-worker-cache-api.https.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226906 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoWrap CDMFairPlayStreaming and related classes in ENABLE(ENCRYPTED_MEDIA) checks
jer.noble@apple.com [Fri, 12 Jan 2018 20:01:02 +0000 (20:01 +0000)]
Wrap CDMFairPlayStreaming and related classes in ENABLE(ENCRYPTED_MEDIA) checks
https://bugs.webkit.org/show_bug.cgi?id=181602

Reviewed by Maciej Stachowiak.

* platform/graphics/avfoundation/CDMFairPlayStreaming.h:
* platform/graphics/avfoundation/objc/CDMInstanceFairPlayStreamingAVFObjC.h:
* platform/graphics/avfoundation/objc/CDMInstanceFairPlayStreamingAVFObjC.mm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226905 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoWebProcess should pass the registration identifier and not the worker identifier...
commit-queue@webkit.org [Fri, 12 Jan 2018 19:43:35 +0000 (19:43 +0000)]
WebProcess should pass the registration identifier and not the worker identifier for fetch events
https://bugs.webkit.org/show_bug.cgi?id=181591

Patch by Youenn Fablet <youenn@apple.com> on 2018-01-12
Reviewed by Chris Dumez.

Source/WebCore:

Test: http/wpt/service-workers/update-service-worker.https.html

Store service worker registration identifier in ResourceLoaderOptions instead of service worker identifier.

* loader/DocumentThreadableLoader.cpp:
(WebCore::DocumentThreadableLoader::DocumentThreadableLoader):
(WebCore::DocumentThreadableLoader::makeCrossOriginAccessRequest):
* loader/ResourceLoaderOptions.h:
* loader/WorkerThreadableLoader.cpp:
(WebCore::WorkerThreadableLoader::MainThreadBridge::MainThreadBridge):
* loader/cache/CachedResourceLoader.cpp:
(WebCore::CachedResourceLoader::prepareFetch):
(WebCore::CachedResourceLoader::determineRevalidationPolicy const):
* loader/cache/CachedResourceRequest.cpp:
(WebCore::CachedResourceRequest::setSelectedServiceWorkerRegistrationIdentifierIfNeeded):
(WebCore::CachedResourceRequest::setNavigationServiceWorkerRegistrationData):
(WebCore::CachedResourceRequest::setSelectedServiceWorkerIdentifierIfNeeded): Deleted.
* loader/cache/CachedResourceRequest.h:
* workers/WorkerScriptLoader.cpp:
(WebCore::WorkerScriptLoader::loadSynchronously):
(WebCore::WorkerScriptLoader::loadAsynchronously):
* workers/service/server/SWServer.cpp:
(WebCore::SWServer::activeWorkerFromRegistrationID):
* workers/service/server/SWServer.h:

Source/WebKit:

Use service worker registration identifier to compute the active service worker identifier responsible to handle the fetch event.

* StorageProcess/ServiceWorker/WebSWServerConnection.cpp:
(WebKit::WebSWServerConnection::startFetch):
* StorageProcess/ServiceWorker/WebSWServerConnection.h:
* StorageProcess/ServiceWorker/WebSWServerConnection.messages.in:
* WebProcess/Storage/ServiceWorkerClientFetch.cpp:
(WebKit::ServiceWorkerClientFetch::start):
* WebProcess/Storage/WebSWClientConnection.cpp:
(WebKit::WebSWClientConnection::startFetch):
* WebProcess/Storage/WebSWClientConnection.h:
* WebProcess/Storage/WebServiceWorkerProvider.cpp:
(WebKit::shouldHandleFetch):

LayoutTests:

* http/wpt/service-workers/resources/empty.html: Added.
* http/wpt/service-workers/update-service-worker.https-expected.txt: Added.
* http/wpt/service-workers/update-service-worker.https.html: Added.
* http/wpt/service-workers/update-worker.py: Added.
(main):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226904 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoUnreviewed, rolling out r226826.
ryanhaddad@apple.com [Fri, 12 Jan 2018 19:21:00 +0000 (19:21 +0000)]
Unreviewed, rolling out r226826.

Breaks internal builds.

Reverted changeset:

"Implement MultiDocument protocol for restoring focus to a
WKWebView"
https://bugs.webkit.org/show_bug.cgi?id=181510
https://trac.webkit.org/changeset/226826

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226903 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agofast/events/ios/rotation/layout-viewport-during-safari-type-rotation.html is flakey
simon.fraser@apple.com [Fri, 12 Jan 2018 19:19:07 +0000 (19:19 +0000)]
fast/events/ios/rotation/layout-viewport-during-safari-type-rotation.html is flakey
https://bugs.webkit.org/show_bug.cgi?id=181569
rdar://problem/34117680

Reviewed by Zalan Bujtas.

Wait a tick before starting the rotation, otherwise -[WKWebView _beginAnimatedResizeWithUpdates:]
can bail because the unobscuredRect is empty.

Also wait for both the rotation UI script to complete, and the receipt of the
orientationchange event.

* fast/events/ios/rotation/layout-viewport-during-safari-type-rotation.html:
* fast/events/ios/rotation/resources/rotation-utils.js:
(doTest):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226902 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[Win][CMake] Remove all-in-one file for WebCore DerivedSources
commit-queue@webkit.org [Fri, 12 Jan 2018 19:05:43 +0000 (19:05 +0000)]
[Win][CMake] Remove all-in-one file for WebCore DerivedSources
https://bugs.webkit.org/show_bug.cgi?id=181582

Patch by Fujii Hironori <Hironori.Fujii@sony.com> on 2018-01-12
Reviewed by Alex Christensen.

Those source files are compiled in unified source build nowadays.

No new tests (No behavior change)

* CMakeLists.txt: Removed calling PROCESS_ALLINONE_FILE.
* DerivedSources.cpp: Removed.
* WebCoreMacros.cmake: Removed a macro PROCESS_ALLINONE_FILE.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226901 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[WinCairo][CMake] Use ${CURL_LIBRARY} instead of specifying the library name libcurl_...
commit-queue@webkit.org [Fri, 12 Jan 2018 18:52:26 +0000 (18:52 +0000)]
[WinCairo][CMake] Use ${CURL_LIBRARY} instead of specifying the library name libcurl_imp explicitly
https://bugs.webkit.org/show_bug.cgi?id=181578

Patch by Fujii Hironori <Hironori.Fujii@sony.com> on 2018-01-12
Reviewed by Alex Christensen.

vcpkg has libcurl.lib, not libcurl_imp.lib.

No new tests because no behavior change.

* PlatformWinCairo.cmake: Link ${CURL_LIBRARY} instead of libcurl_imp.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226900 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[WK2] Unify macOS and iOS codepaths in the UI process when starting a drag
wenson_hsieh@apple.com [Fri, 12 Jan 2018 18:49:11 +0000 (18:49 +0000)]
[WK2] Unify macOS and iOS codepaths in the UI process when starting a drag
https://bugs.webkit.org/show_bug.cgi?id=181574

Reviewed by Tim Horton.

Rearrange some macOS drag start logic in the UI process so that it uses the same codepaths in WebPageProxy as
iOS. Namely, startDrag should just forward the DragItem and drag image handle along to the appropriate views on
each platform (WKContentView and WebViewImpl).

There should be no change in behavior.

* UIProcess/Cocoa/WebPageProxyCocoa.mm:

Both macOS and iOS now funnel through this method.

* UIProcess/Cocoa/WebViewImpl.h:
* UIProcess/Cocoa/WebViewImpl.mm:
(WebKit::WebViewImpl::startDrag):
(WebKit::WebViewImpl::dragImageForView): Deleted.

Rename dragImageForView to startDrag. Move the call to didStartDrag() here, and call dragCancelled() in the
case where we bailed from starting the drag (due to failing to create a drag image).

* UIProcess/PageClient.h:
* UIProcess/ios/PageClientImplIOS.h:
* UIProcess/ios/PageClientImplIOS.mm:
(WebKit::PageClientImpl::setDragImage): Deleted.

Rename setDragImage to startDrag.

* UIProcess/mac/PageClientImplMac.h:
* UIProcess/mac/PageClientImplMac.mm:
(WebKit::PageClientImpl::startDrag):
(WebKit::PageClientImpl::setDragImage): Deleted.
* UIProcess/mac/WebPageProxyMac.mm:
(WebKit::WebPageProxy::startDrag): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226899 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoWebGL video texture black in Safari 11.0.2 and wrong colored in Safari Preview 11.1
jer.noble@apple.com [Fri, 12 Jan 2018 18:30:06 +0000 (18:30 +0000)]
WebGL video texture black in Safari 11.0.2 and wrong colored in Safari Preview 11.1
https://bugs.webkit.org/show_bug.cgi?id=181445
<rdar://problem/36383183>

Reviewed by Dean Jackson.

Tests: added compile-time correctness tests for YCbCrMatrix values.

Perform the derivation from YCbCr coefficients into matrices in constexpr expressions
at compile-time. This allows us to also perform compile-time correctness checks to catch
regressions which may cause incorrect color conversions.

Since we now have general-purpose derivation of matrix values from coefficients,
adding missing specificed matrices is trivial, so add support for SMPTE 240M and BT.2020
matrices.

* platform/graphics/cv/VideoTextureCopierCV.cpp:
(WebCore::GLfloatColor::GLfloatColor):
(WebCore::GLfloatColor::abs):
(WebCore::GLfloatColor::isApproximatelyEqualTo const):
(WebCore::YCbCrMatrix::operator Vector<GLfloat> const):
(WebCore::YCbCrMatrix::YCbCrMatrix):
(WebCore::YCbCrMatrix::operator* const):
(WebCore::YCbCrToRGBMatrixForRangeAndTransferFunction):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226898 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoREGRESSION (r226818): API test WebKitLegacy.LoadInvalidURLRequest is failing
achristensen@apple.com [Fri, 12 Jan 2018 18:26:53 +0000 (18:26 +0000)]
REGRESSION (r226818): API test WebKitLegacy.LoadInvalidURLRequest is failing
https://bugs.webkit.org/show_bug.cgi?id=181595

This test was testing behavior of an invalid URL that WebCore parses but NSURL does not.
It was using example.com<> but after r226479 < and > are considered invalid by WebCore.
I change the < and > to $ to make this test pass and continue to test what it used to.

* TestWebKitAPI/Tests/mac/LoadInvalidURLRequest.html:
* TestWebKitAPI/Tests/mac/LoadInvalidURLRequest.mm:
(-[LoadInvalidURLWebFrameLoadDelegate webView:didFailProvisionalLoadWithError:forFrame:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226897 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoSkipped http/tests/resourceLoadStatistics/grandfathering.html on macOS WK2.
jlewis3@apple.com [Fri, 12 Jan 2018 17:39:49 +0000 (17:39 +0000)]
Skipped http/tests/resourceLoadStatistics/grandfathering.html on macOS WK2.
https://bugs.webkit.org/show_bug.cgi?id=181482

Unreviewed test gardening.

* platform/mac-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226896 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoWeb Inspector: Remove unnecessary raw pointer in InspectorConsoleAgent
commit-queue@webkit.org [Fri, 12 Jan 2018 17:32:56 +0000 (17:32 +0000)]
Web Inspector: Remove unnecessary raw pointer in InspectorConsoleAgent
https://bugs.webkit.org/show_bug.cgi?id=181579
<rdar://problem/36193759>

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2018-01-12
Reviewed by Brian Burg.

* inspector/agents/InspectorConsoleAgent.h:
* inspector/agents/InspectorConsoleAgent.cpp:
(Inspector::InspectorConsoleAgent::clearMessages):
(Inspector::InspectorConsoleAgent::addConsoleMessage):
Switch from a raw pointer to m_consoleMessages.last().
Also move the expiration check into the if block since it can only
happen inside here when the number of console messages changes.

(Inspector::InspectorConsoleAgent::discardValues):
Also clear the expired message count when messages are cleared.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226895 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoAdd support for the frames() timing function
graouts@webkit.org [Fri, 12 Jan 2018 13:26:50 +0000 (13:26 +0000)]
Add support for the frames() timing function
https://bugs.webkit.org/show_bug.cgi?id=181585
<rdar://problem/36463317>

Reviewed by Dean.

Source/WebCore:

Implement the frames() timing function as specified in the CSS Timing Functions Level 1
specification, specifically https://www.w3.org/TR/css-timing-1/#frames-timing-functions.
A frames timing function is a type of timing function that divides the input time into a
specified number of intervals of equal length.

Test: transitions/frames-timing-function.html

* css/CSSComputedStyleDeclaration.cpp:
(WebCore::createTimingFunctionValue):
* css/CSSTimingFunctionValue.cpp:
(WebCore::CSSFramesTimingFunctionValue::customCSSText const):
(WebCore::CSSFramesTimingFunctionValue::equals const):
* css/CSSTimingFunctionValue.h:
* css/CSSToStyleMap.cpp:
(WebCore::CSSToStyleMap::mapAnimationTimingFunction):
* css/CSSValue.cpp:
(WebCore::CSSValue::equals const):
(WebCore::CSSValue::cssText const):
(WebCore::CSSValue::destroy):
* css/CSSValue.h:
(WebCore::CSSValue::isFramesTimingFunctionValue const):
* css/CSSValueKeywords.in:
* css/parser/CSSPropertyParser.cpp:
(WebCore::consumeSteps):
(WebCore::consumeFrames):
(WebCore::consumeAnimationTimingFunction):
* platform/animation/TimingFunction.cpp:
(WebCore::operator<<):
(WebCore::TimingFunction::transformTime const):
* platform/animation/TimingFunction.h:
(WebCore::TimingFunction::isFramesTimingFunction const):
* platform/graphics/ca/GraphicsLayerCA.cpp:
(WebCore::animationHasFramesTimingFunction):
(WebCore::GraphicsLayerCA::animationCanBeAccelerated const):

Source/WebKit:

Add the ability to endode and decode the frames() timing function.

* Shared/WebCoreArgumentCoders.cpp:
(IPC::ArgumentCoder<FramesTimingFunction>::encode):
(IPC::ArgumentCoder<FramesTimingFunction>::decode):
* Shared/WebCoreArgumentCoders.h:
* WebProcess/WebPage/RemoteLayerTree/PlatformCAAnimationRemote.mm:
(WebKit::PlatformCAAnimationRemote::Properties::encode const):
(WebKit::PlatformCAAnimationRemote::Properties::decode):

LayoutTests:

Add a new test that checks that the frames() timing function applies as expected
and expand an existing test to check that the frames() timing function is parsed
correctly. We also mark progressions in imported WPT tests.

* imported/w3c/web-platform-tests/css-timing-1/frames-timing-functions-output-expected.txt:
* imported/w3c/web-platform-tests/css-timing-1/frames-timing-functions-syntax-expected.txt:
* transitions/frames-timing-function-expected.txt: Added.
* transitions/frames-timing-function.html: Added.
* transitions/transitions-parsing-expected.txt:
* transitions/transitions-parsing.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226886 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[JSC] Create parallel SlotVisitors apriori
utatane.tea@gmail.com [Fri, 12 Jan 2018 12:16:12 +0000 (12:16 +0000)]
[JSC] Create parallel SlotVisitors apriori
https://bugs.webkit.org/show_bug.cgi?id=180907

Reviewed by Saam Barati.

The number of SlotVisitors are capped with the number of HeapHelperPool's threads + 2.
If we create these SlotVisitors apropri, we do not need to create SlotVisitors dynamically.
Then we do not need to grab locks while iterating all the SlotVisitors.

In addition, we do not need to consider the case that the number of SlotVisitors increases
after setting up VisitCounters in MarkingConstraintSolver since the number of SlotVisitors
does not increase any more.

* heap/Heap.cpp:
(JSC::Heap::Heap):
(JSC::Heap::runBeginPhase):
* heap/Heap.h:
* heap/HeapInlines.h:
(JSC::Heap::forEachSlotVisitor):
(JSC::Heap::numberOfSlotVisitors): Deleted.
* heap/MarkingConstraintSolver.cpp:
(JSC::MarkingConstraintSolver::didVisitSomething const):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226885 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoEach variant of a polymorphic inlined call should be exitOK at the top of the block
sbarati@apple.com [Fri, 12 Jan 2018 10:15:09 +0000 (10:15 +0000)]
Each variant of a polymorphic inlined call should be exitOK at the top of the block
https://bugs.webkit.org/show_bug.cgi?id=181562
<rdar://problem/36445624>

Reviewed by Yusuke Suzuki.

JSTests:

* stress/each-block-at-top-of-polymorphic-call-inlining-should-be-exitOK.js: Added.
(f):
(foo):

Source/JavaScriptCore:

Before this patch, the very first block in the switch for polymorphic call
inlining will have exitOK at the top. The others are not guaranteed to.
That was just a bug. They're all exitOK at the top. This will lead to crashes
in FixupPhase because we won't have a node in a block that has ExitOK, so
when we fixup various type checks, we assert out.

* dfg/DFGByteCodeParser.cpp:
(JSC::DFG::ByteCodeParser::handleInlining):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226881 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoUnreviewed, rolling out r226721.
commit-queue@webkit.org [Fri, 12 Jan 2018 08:39:54 +0000 (08:39 +0000)]
Unreviewed, rolling out r226721.
https://bugs.webkit.org/show_bug.cgi?id=181583

Lets do a slightly different fix (Requested by anttik on
#webkit).

Reverted changeset:

"REGRESSION(r225650): The scores of MotionMark tests Multiply
and Leaves dropped by 8%"
https://bugs.webkit.org/show_bug.cgi?id=181460
https://trac.webkit.org/changeset/226721

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226880 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoRename ENABLE_ASYNC_ITERATION to ENABLE_JS_ASYNC_ITERATION
keith_miller@apple.com [Fri, 12 Jan 2018 07:59:30 +0000 (07:59 +0000)]
Rename ENABLE_ASYNC_ITERATION to ENABLE_JS_ASYNC_ITERATION
https://bugs.webkit.org/show_bug.cgi?id=181573

Reviewed by Simon Fraser.

.:

* Source/cmake/WebKitFeatures.cmake:

Source/JavaScriptCore:

* Configurations/FeatureDefines.xcconfig:
* runtime/Options.h:

Source/WebCore:

* Configurations/FeatureDefines.xcconfig:

Source/WebCore/PAL:

* Configurations/FeatureDefines.xcconfig:

Source/WebKit:

* Configurations/FeatureDefines.xcconfig:

Source/WebKitLegacy/mac:

* Configurations/FeatureDefines.xcconfig:

Tools:

* TestWebKitAPI/Configurations/FeatureDefines.xcconfig:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226879 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoRemove noexcept from definition of std::tie()
commit-queue@webkit.org [Fri, 12 Jan 2018 06:53:51 +0000 (06:53 +0000)]
Remove noexcept from definition of std::tie()
https://bugs.webkit.org/show_bug.cgi?id=181577

Patch by Basuke Suzuki <Basuke.Suzuki@sony.com> on 2018-01-11
Reviewed by Yusuke Suzuki.

* wtf/StdLibExtras.h:
(WTF::tie):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226878 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[WebAuthN] Import a CBOR coder from Chromium
jiewen_tan@apple.com [Fri, 12 Jan 2018 05:29:01 +0000 (05:29 +0000)]
[WebAuthN] Import a CBOR coder from Chromium
https://bugs.webkit.org/show_bug.cgi?id=181522
<rdar://problem/36055729>

Reviewed by Brent Fulgham.

Source/WebCore:

This patch imports a CBOR coder including encoder and decoder from Chromium. CBOR encoder
is needed for WebAuthN to encode attestation object into binaries. When supporting extensions
in the future, CBOR encoder/decoder will be needed as well. Implementating and maintaining
a fully-fledged CBOR coder doesn't seem to align with WebKit's best interests. Therefore,
importing a most suitable third party implementation will be wise.

In this patch, it fully integrate the whole coder into our codebase. Those changes includes:
    1. Substitute data structures that enjoy a better WTF version.
    2. Replacing marcos.
    3. Implementating workarounds for some functionalities that we lack fundamental types' support.
    4. Changing the coding style to match ours.

This patch doesn't intend to improve the logic of the original codebase. Hence some of the
coding logic might not match what WebKit ususally has.

Here is a full list of Chromium changes that constructed this CBOR coder in chronological order:
6efcf495521d18d060027762f48bb292d6979136,
9eb43fd347890b4c6cf54c4bd7ec1bbb88e381e1,
31c85e74fd567772f18e0a41be468d04af721f21,
68672fdcad280a8ff69b91927d38d0eabf2c87f2,
0ca8667c0584fb21c0748ebd7468d32889759a07,
df763d790d7e45d70116bdefacbfd4f9faa8995e,
6d30c4a621c65314db63eb56e87c19ab75627b26,
50fe92953f4739f17a62303fedbf8db9234317c8,
47be22c3603424d1832d046a348ff3f982500288,
98a59e46948b2c71608926004fac8192b0ff2208,
07540c6d850ed6e0fa508d63c20a8ce96d751de6,
06ae32d640c8e4b86ea8914a80ee419ea16e56d8.

Covered by API tests.

* Modules/webauthn/cbor/CBORBinary.h: Added.
* Modules/webauthn/cbor/CBORReader.cpp: Added.
(cbor::CBORReader::CBORReader):
(cbor::CBORReader::~CBORReader):
(cbor::CBORReader::read):
(cbor::CBORReader::decodeCBOR):
(cbor::CBORReader::readVariadicLengthInteger):
(cbor::CBORReader::decodeValueToNegative):
(cbor::CBORReader::decodeValueToUnsigned):
(cbor::CBORReader::readSimpleValue):
(cbor::CBORReader::readString):
Workarounds applied.
(cbor::CBORReader::readBytes):
(cbor::CBORReader::readCBORArray):
(cbor::CBORReader::readCBORMap):
(cbor::CBORReader::canConsume):
(cbor::CBORReader::checkMinimalEncoding):
(cbor::CBORReader::checkExtraneousData):
(cbor::CBORReader::checkDuplicateKey):
(cbor::CBORReader::hasValidUTF8Format):
Workarounds applied.
(cbor::CBORReader::checkOutOfOrderKey):
(cbor::CBORReader::getErrorCode):
(cbor::CBORReader::errorCodeToString):
* Modules/webauthn/cbor/CBORReader.h: Added.
* Modules/webauthn/cbor/CBORValue.cpp: Added.
(cbor::CBORValue::CBORValue):
(cbor::CBORValue::operator=):
(cbor::CBORValue::~CBORValue):
(cbor::CBORValue::clone const):
(cbor::CBORValue::getInteger const):
(cbor::CBORValue::getUnsigned const):
(cbor::CBORValue::getNegative const):
(cbor::CBORValue::getString const):
(cbor::CBORValue::getByteString const):
(cbor::CBORValue::getArray const):
(cbor::CBORValue::getMap const):
(cbor::CBORValue::getSimpleValue const):
(cbor::CBORValue::internalMoveConstructFrom):
(cbor::CBORValue::internalCleanup):
* Modules/webauthn/cbor/CBORValue.h: Added.
* Modules/webauthn/cbor/CBORWriter.cpp: Added.
(cbor::CBORWriter::~CBORWriter):
(cbor::CBORWriter::write):
(cbor::CBORWriter::CBORWriter):
(cbor::CBORWriter::encodeCBOR):
Workarounds applied.
(cbor::CBORWriter::startItem):
(cbor::CBORWriter::setAdditionalInformation):
(cbor::CBORWriter::setUint):
(cbor::CBORWriter::getNumUintBytes):
* Modules/webauthn/cbor/CBORWriter.h: Added.
* Sources.txt:
* WebCore.xcodeproj/project.pbxproj:

Tools:

This patch also imports all unit tests into our API tests to ensure all
workarounds and modification against the original codebase doesn't change
any original functionalities.

* TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
* TestWebKitAPI/Tests/WebCore/CBORReaderTest.cpp: Added.
(TestWebKitAPI::TEST):
* TestWebKitAPI/Tests/WebCore/CBORValueTest.cpp: Added.
(TestWebKitAPI::TEST):
* TestWebKitAPI/Tests/WebCore/CBORWriterTest.cpp: Added.
(TestWebKitAPI::eq):
Workarounds applied.
(TestWebKitAPI::TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226862 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoSetting Window.opener to null should disown its opener
cdumez@apple.com [Fri, 12 Jan 2018 04:47:31 +0000 (04:47 +0000)]
Setting Window.opener to null should disown its opener
https://bugs.webkit.org/show_bug.cgi?id=181505
<rdar://problem/36443151>

Reviewed by Ryosuke Niwa.

Source/WebCore:

Setting Window.opener to null should disown its opener as per:
- https://html.spec.whatwg.org/#dom-opener

With this change, tabs opened by clicking link inside Gmail no
longer have the Gmail window as opener.

Tests: fast/dom/Window/window-opener-set-to-null.html
       fast/dom/Window/window-opener-shadowing.html

* bindings/js/JSDOMWindowCustom.cpp:
(WebCore::JSDOMWindow::setOpener):
* page/DOMWindow.cpp:
(WebCore::DOMWindow::disownOpener):
* page/DOMWindow.h:
* page/DOMWindow.idl:

LayoutTests:

Add layout test coverage.

* fast/dom/Window/window-opener-set-to-null-expected.txt: Added.
* fast/dom/Window/window-opener-set-to-null.html: Added.
* fast/dom/Window/window-opener-shadowing-expected.txt: Added.
* fast/dom/Window/window-opener-shadowing.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226842 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agowebkitpy: Use partially disambiguated type in SimulatedDeviceManager._disambiguate_de...
jbedard@apple.com [Fri, 12 Jan 2018 03:31:27 +0000 (03:31 +0000)]
webkitpy: Use partially disambiguated type in SimulatedDeviceManager._disambiguate_device_type
https://bugs.webkit.org/show_bug.cgi?id=181538
<rdar://problem/36440580>

Reviewed by Aakash Jain.

When disambiguating a device type, it is possible that the provided device type is sufficiently
ambiguous that the comparisons against complete device types without a disambiguated hardware_family
will result in a failure to disambiguate the type.

* Scripts/webkitpy/xcode/new_simulated_device.py:
(SimulatedDeviceManager._disambiguate_device_type):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226841 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoREGRESSION(226788): AppStore Crashed @ JavaScriptCore: JSC::MacroAssemblerARM64:...
msaboff@apple.com [Fri, 12 Jan 2018 03:30:40 +0000 (03:30 +0000)]
REGRESSION(226788): AppStore Crashed @ JavaScriptCore: JSC::MacroAssemblerARM64::pushToSaveImmediateWithoutTouchingRegisters
https://bugs.webkit.org/show_bug.cgi?id=181570

Reviewed by Keith Miller.

* assembler/MacroAssemblerARM64.h:
(JSC::MacroAssemblerARM64::abortWithReason):
Reverting these functions to use dataTempRegister and memoryTempRegister as they are
JIT release asserts that will crash the program.

(JSC::MacroAssemblerARM64::pushToSaveImmediateWithoutTouchingRegisters):
Changed this so that it invalidates any cached dataTmpRegister contents if temp register
caching is enabled.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226840 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months ago[Cocoa] Star character disappears when bolded
mmaxfield@apple.com [Fri, 12 Jan 2018 03:16:52 +0000 (03:16 +0000)]
[Cocoa] Star character disappears when bolded
https://bugs.webkit.org/show_bug.cgi?id=181568
<rdar://problem/18755569>

Reviewed by Simon Fraser.

Source/WebCore:

We had some code in ComplexTextController to ask the FontCache for a font, given the name
of a particular font CoreText used inside a CTRun. This is wrong for two reasons: fonts
are not identifiable by PostScript name (in the general case), and because the lookup
procedure requires a FontDescription, the result may yield a font that is not the one
looked up. The goal of this code was simply to preserve the rendering mode of the font,
but we removed support for these rendering modes years ago. So the solution is to skip
that lookup and use the CoreText font directly.

Test: fast/text/unknown-font.html

* platform/graphics/mac/ComplexTextControllerCoreText.mm:
(WebCore::ComplexTextController::collectComplexTextRunsForCharacters):

LayoutTests:

* fast/text/unknown-font-expected-mismatch.html: Added.
* fast/text/unknown-font.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226839 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoREGRESSION(r219530): ResourceLoadStatisticsPersistentStorage should be read-only...
bfulgham@apple.com [Fri, 12 Jan 2018 03:13:05 +0000 (03:13 +0000)]
REGRESSION(r219530): ResourceLoadStatisticsPersistentStorage should be read-only in ephemeral sessions
https://bugs.webkit.org/show_bug.cgi?id=181136
<rdar://problem/36116604>

Reviewed by Chris Dumez.

Source/WebKit:

Some uses of WebKit involve running a UIProcess as an ephemeral session for the life of the process. In this
case, we do not initialize the data path for the set of load statistics triggering an assertion.

We actually intended ephemeral sessions to consume the existing resource load data (presumably captured during
non-ephemeral browsing). This would be a read-only mode, where it would not add new entries to the load
statistics, but would take advantage of existing observations. Currently that does not happen (for this type
of WebKit embed), which forces each run as an ephemeral session to build up in-memory browsing data until it has
enough observations to begin modifying loads.

We need to set the ResourceLoadStatisticsPersistentStorage object to a "read only" mode in this case, so
that it read (but does not write) from this database.

Tested by ephemeral website data TestWebKitAPI tests.

* UIProcess/ResourceLoadStatisticsPersistentStorage.cpp:
(WebKit::ResourceLoadStatisticsPersistentStorage::create): Added to allow creation of the right style of
Persistent Storage.
(WebKit::ResourceLoadStatisticsPersistentStorage::ResourceLoadStatisticsPersistentStorage): Initialize the
new data member.
(WebKit::ResourceLoadStatisticsPersistentStorage::asyncWriteTimerFired): RELEASE_ASSERT that we never run
this method when in "read only" mode.
(WebKit::ResourceLoadStatisticsPersistentStorage::writeMemoryStoreToDisk): Ditto.
(WebKit::ResourceLoadStatisticsPersistentStorage::scheduleOrWriteMemoryStore): Return early if asked to
schedule a write operation for a "read only" persistent store.
(WebKit::ResourceLoadStatisticsPersistentStorage::finishAllPendingWorkSynchronously): RELEASE_ASSERT if we
ever shut down in "read only" mode with an active write timer.
* UIProcess/ResourceLoadStatisticsPersistentStorage.h:
* UIProcess/WebResourceLoadStatisticsStore.cpp:
(WebKit::WebResourceLoadStatisticsStore::WebResourceLoadStatisticsStore): Pass a flag indicating whether the
storage session is ephemeral or not.
* UIProcess/WebResourceLoadStatisticsStore.h:

Tools:

Add a new API test to confirm that ResourceLoadStatistics can be turned on safely for ephemeral
browsing sessions.

* Scripts/run-gtk-tests:
(GtkTestRunner): Unskip test now that it passes.
* TestWebKitAPI/Tests/WebKitCocoa/WebsiteDataStoreCustomPaths.mm:
(TEST): Add new WebsiteDataStoreEphemeral test.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226838 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoUnreviewed, rolling out r226816.
ryanhaddad@apple.com [Fri, 12 Jan 2018 02:19:25 +0000 (02:19 +0000)]
Unreviewed, rolling out r226816.

This change broke LayoutTests on the bots.

Reverted changeset:

"run-webkit-tests fails when there is a curly brace in Xcode
build output"
https://bugs.webkit.org/show_bug.cgi?id=181254
https://trac.webkit.org/changeset/226816

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226837 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoCannot trigger Dromaeo tests on internal perf try bots
rniwa@webkit.org [Fri, 12 Jan 2018 02:13:53 +0000 (02:13 +0000)]
Cannot trigger Dromaeo tests on internal perf try bots
https://bugs.webkit.org/show_bug.cgi?id=179712

Reviewed by Chris Dumez.

The bug was caused by CustomAnalysisTaskConfigurator only showing the top-level tests that are triggerable
instead of the list of highest level tests that are triggerable.

* public/v3/components/custom-analysis-task-configurator.js:
(CustomAnalysisTaskConfigurator.prototype.selectTests): Update the test group name when a new test is picked.
(CustomAnalysisTaskConfigurator.prototype.selectPlatform):
(CustomAnalysisTaskConfigurator.prototype._didUpdateSelectedPlatforms): Extracted from selectPlatform.
(CustomAnalysisTaskConfigurator.prototype._renderTriggerableTests): Include the list of all highest-level tests
which are triggerable.
(CustomAnalysisTaskConfigurator.prototype._renderRadioButtonList): Added labelForObject which returns the label
to be used in the list items. For tests, we want to use the full name, not just its label.
* public/v3/models/analysis-task.js:
(AnalysisTask.fetchById):
* public/v3/models/triggerable.js:
(Triggerable.prototype.acceptedTests): Added.
(Triggerable.prototype.acceptsTest): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226836 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoAdd optional logging of ITP-related user interaction information
krollin@apple.com [Fri, 12 Jan 2018 01:53:58 +0000 (01:53 +0000)]
Add optional logging of ITP-related user interaction information
https://bugs.webkit.org/show_bug.cgi?id=181556

Reviewed by Brent Fulgham.

In order to support the tracking of the efficacy of Intelligent
Tracking Protection, add some logging indicating when the user
interacts with a page in a way that affects cookie partitioning. This
logging is off by default, and is enabled with `defaults write -g
WebKitLogCookieInformation -bool true`.

Source/WebCore:

No new tests -- no changed functionality.

* loader/ResourceLoadObserver.cpp:
(WebCore::ResourceLoadObserver::logUserInteractionWithReducedTimeResolution):
* loader/ResourceLoadObserver.h:
(WebCore::ResourceLoadObserver::shouldLogUserInteraction const):
(WebCore::ResourceLoadObserver::setShouldLogUserInteraction):

Source/WebKit:

* Shared/WebProcessCreationParameters.cpp:
(WebKit::WebProcessCreationParameters::encode const):
(WebKit::WebProcessCreationParameters::decode):
* Shared/WebProcessCreationParameters.h:
* UIProcess/Cocoa/WebProcessPoolCocoa.mm:
(WebKit::WebProcessPool::platformInitializeWebProcess):
* WebProcess/WebProcess.cpp:
(WebKit::WebProcess::initializeWebProcess):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226835 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoImplement MultiDocument protocol for restoring focus to a WKWebView
megan_gardner@apple.com [Fri, 12 Jan 2018 01:43:06 +0000 (01:43 +0000)]
Implement MultiDocument protocol for restoring focus to a WKWebView
https://bugs.webkit.org/show_bug.cgi?id=181510

Reviewed by Dan Bernstein.

Support the UIKit protocol for restoring focus to a what previously had focus.
WebKit already has a method to silently remove and replace focus, without telling the
web process about the unfocus and refocusing, so we're just using that.

* Platform/spi/ios/UIKitSPI.h:
* UIProcess/ios/WKContentViewInteraction.h:
* UIProcess/ios/WKContentViewInteraction.mm:
(-[WKContentView _restoreFocusWithToken:]):
(-[WKContentView _preserveFocusWithToken:destructively:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226826 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoAX: when invert colors is on, double-invert video elements in UserAgentStyleSheet
jcraig@apple.com [Fri, 12 Jan 2018 01:16:59 +0000 (01:16 +0000)]
AX: when invert colors is on, double-invert video elements in UserAgentStyleSheet
https://bugs.webkit.org/show_bug.cgi?id=168447
<rdar://problem/30559874>

Reviewed by Simon Fraser.

Double-invert video when platform "invert colors" setting is enabled. Behavior matches
current "Smart Invert" feature of Safari Reader on macOS/iOS and other iOS native apps.

Source/WebCore:

Tests: accessibility/smart-invert-reference.html
       accessibility/smart-invert.html

* Modules/modern-media-controls/controls/media-controls.css:
(@media (inverted-colors)):
(:host):
(picture):
* css/html.css:
(@media (inverted-colors)):
(video):

LayoutTests:

* TestExpectations: Platform setting only available on Mac and iOS.
* accessibility/smart-invert-expected.txt: Added.
* accessibility/smart-invert-reference-expected.html: Added.
* accessibility/smart-invert-reference.html: Added. Ref to ensure invert and grayscale filters render as expected.
* accessibility/smart-invert.html: Added. Computed expectatons of filter property text values.
* platform/ios-wk2/TestExpectations: Runs on iOS WK2.
* platform/mac-wk2/TestExpectations: Runs on Mac WK2.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226825 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoDon't load inline data when requesting info for an attachment element backed by a...
wenson_hsieh@apple.com [Fri, 12 Jan 2018 01:13:28 +0000 (01:13 +0000)]
Don't load inline data when requesting info for an attachment element backed by a file path
https://bugs.webkit.org/show_bug.cgi?id=181550

Source/WebCore:

Reviewed by Tim Horton.

When requesting data for an attachment element that is backed by a file path, we currently trigger a load in the
web process to fetch contents of the attachment data as inline data in the AttachmentInfo. This is unnecessary,
since the file path of the attachment element must have come from the UI process anyways, so it is sufficient to
simply send the file path to the UI process and have the UI process read the contents of the path as a memory-
mapped NSData.

This patch lets HTMLAttachmentElement skip over resource loading codepaths when creating an AttachmentInfo for
the client, and also teaches _WKAttachment to read a AttachmentInfo's filepath as memory-mapped data if a file
path is present, and no inline data was specified.

Covered by existing API tests.

* html/HTMLAttachmentElement.cpp:
(WebCore::HTMLAttachmentElement::requestInfo):

Source/WebKit:

Reviewed by Tim Horton

See WebCore/ChangeLog for more information.

* UIProcess/API/Cocoa/_WKAttachment.mm:
(-[_WKAttachmentInfo initWithInfo:]):
(-[_WKAttachmentInfo fileLoadingError]):
(-[_WKAttachment requestInfo:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226824 268f45cc-cd09-0410-ab3c-d52691b4dbfc

21 months agoMake elements of zero width or height focusable
rniwa@webkit.org [Fri, 12 Jan 2018 00:56:59 +0000 (00:56 +0000)]
Make elements of zero width or height focusable
https://bugs.webkit.org/show_bug.cgi?id=181516

Reviewed by Chris Dumez.

Source/WebCore:

Don't check render box's size or bounding rect when deciding whether an element is focusable.
New behavior matches that of Firefox and Chrome.

Test: fast/events/focus-zero-size-element.html

* dom/Element.cpp:
(WebCore::Element::isFocusable): Only update the style.
* html/HTMLFormControlElement.cpp:
(WebCore::HTMLFormControlElement::isFocusable const): Deleted.
* html/HTMLFormControlElement.h:
* mathml/MathMLElement.cpp:
(WebCore::MathMLElement::isFocusable const): Deleted. As far as I can tell, no math ml element is focusable.
* mathml/MathMLElement.h:
* svg/SVGAElement.cpp:
(WebCore::SVGAElement::isFocusable const): Deleted.
* svg/SVGAElement.h:

LayoutTests:

Added a regression test.

* fast/events/focus-zero-size-element-expected.txt: Added.
* fast/events/focus-zero-size-element.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@226823 268f45cc-cd09-0410-ab3c-d52691b4dbfc