WebKit-https.git
3 years agojsc shell's flashHeapAccess() should not do JS work after releasing access to the...
mark.lam@apple.com [Thu, 30 Nov 2017 22:51:59 +0000 (22:51 +0000)]
jsc shell's flashHeapAccess() should not do JS work after releasing access to the heap.
https://bugs.webkit.org/show_bug.cgi?id=180219
<rdar://problem/35696536>

Reviewed by Filip Pizlo.

JSTests:

* stress/regress-180219.js: Added.

Source/JavaScriptCore:

* jsc.cpp:
(functionFlashHeapAccess):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225352 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoServiceWorkerThread's m_extendedEvents gets destroyed on the wrong thread
cdumez@apple.com [Thu, 30 Nov 2017 22:35:25 +0000 (22:35 +0000)]
ServiceWorkerThread's m_extendedEvents gets destroyed on the wrong thread
https://bugs.webkit.org/show_bug.cgi?id=180216

Reviewed by Brady Eidson.

ServiceWorkerThread's m_extendedEvents gets destroyed on the wrong thread. Those events live on the worker
thread so we should destroy them on the worker thread, not the main thread. To address the issue, m_extendedEvents
was moved to ServiceWorkerGlobalScope, which actually lives on the right thread.

* workers/service/ServiceWorkerGlobalScope.cpp:
(WebCore::ServiceWorkerGlobalScope::updateExtendedEventsSet):
* workers/service/ServiceWorkerGlobalScope.h:
* workers/service/context/ServiceWorkerThread.cpp:
(WebCore::ServiceWorkerThread::postFetchTask):
(WebCore::ServiceWorkerThread::postMessageToServiceWorkerGlobalScope):
(WebCore::ServiceWorkerThread::updateExtendedEventsSet): Deleted.
* workers/service/context/ServiceWorkerThread.h:
(WebCore::ServiceWorkerThread::hasPendingEvents const): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225351 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoSkip WebGL on iOS while working out what ANGLE has broken.
dino@apple.com [Thu, 30 Nov 2017 22:25:52 +0000 (22:25 +0000)]
Skip WebGL on iOS while working out what ANGLE has broken.

* platform/ios/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225350 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoMarked http/tests/loading/resourceLoadStatistics/partitioned-and-unpartitioned-cookie...
jlewis3@apple.com [Thu, 30 Nov 2017 22:21:31 +0000 (22:21 +0000)]
Marked http/tests/loading/resourceLoadStatistics/partitioned-and-unpartitioned-cookie-deletion.html as flaky on High Sierra.
https://bugs.webkit.org/show_bug.cgi?id=179644

Unreviewed test gardening.

* platform/mac-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225349 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoMarked imported/w3c/web-platform-tests/webrtc/RTCPeerConnection-onnegotiationneeded...
jlewis3@apple.com [Thu, 30 Nov 2017 21:49:05 +0000 (21:49 +0000)]
Marked imported/w3c/web-platform-tests/webrtc/RTCPeerConnection-onnegotiationneeded.html as flak on macOS.
https://bugs.webkit.org/show_bug.cgi?id=176078

Unreviewed test gardening.

* platform/mac-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225348 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoSWServerToContextConnection / SWServerWorker do not need to be ThreadSafeRefCounted
cdumez@apple.com [Thu, 30 Nov 2017 21:42:29 +0000 (21:42 +0000)]
SWServerToContextConnection / SWServerWorker do not need to be ThreadSafeRefCounted
https://bugs.webkit.org/show_bug.cgi?id=180214

Reviewed by Brady Eidson.

SWServerToContextConnection / SWServerWorker do not need to be ThreadSafeRefCounted.
RefCounted is sufficient as there is no longer any multithreading on server side.

* workers/service/server/SWServerToContextConnection.h:
* workers/service/server/SWServerWorker.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225347 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoServiceWorkerGlobalScope::skipWaiting() should resolve its promise
cdumez@apple.com [Thu, 30 Nov 2017 21:41:59 +0000 (21:41 +0000)]
ServiceWorkerGlobalScope::skipWaiting() should resolve its promise
https://bugs.webkit.org/show_bug.cgi?id=180204

Reviewed by Brady Eidson.

LayoutTests/imported/w3c:

Rebaseline test which uses skipWaiting().

* web-platform-tests/service-workers/service-worker/claim-with-redirect.https-expected.txt:

Source/WebCore:

ServiceWorkerGlobalScope::skipWaiting() should resolve its promise, even if it is not
implemented yet. For now, always reject the promise.

* workers/service/ServiceWorkerGlobalScope.cpp:
(WebCore::ServiceWorkerGlobalScope::skipWaiting):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225346 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoMark a few heap-allocated service worker types as WTF_MAKE_FAST_ALLOCATED
cdumez@apple.com [Thu, 30 Nov 2017 21:16:22 +0000 (21:16 +0000)]
Mark a few heap-allocated service worker types as WTF_MAKE_FAST_ALLOCATED
https://bugs.webkit.org/show_bug.cgi?id=180212

Reviewed by Geoffrey Garen.

Mark a few heap-allocated service worker types as WTF_MAKE_FAST_ALLOCATED for performance.

* workers/service/server/SWOriginStore.h:
* workers/service/server/SWServer.h:
* workers/service/server/SWServerJobQueue.h:
* workers/service/server/SWServerRegistration.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225344 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoMake WorkerThread lifetime much more predictable.
beidson@apple.com [Thu, 30 Nov 2017 21:01:12 +0000 (21:01 +0000)]
Make WorkerThread lifetime much more predictable.
https://bugs.webkit.org/show_bug.cgi?id=180203

Reviewed by Chris Dumez.

No new tests (Fixes flakiness in existing and future tests).

The family of classes related to Workers has a complicated ownership model.

For Dedicated Workers, the WorkerThread object is owned by the WorkerMessagingProxy,
which manages its own lifetime. Additionally, other object(s) have raw C++ references
to it, and the expected lifetimes are described in comments scattered through a few files.

What it boils down to is that the "Worker" DOM object - which lives on the main thread -
is the key to the proper destruction of all of these objects.

For ServiceWorkers running in their own context process, there is no "Worker" on the main thread.

As a result, ServiceWorkers can get into a situation where their WorkerThread can be destroyed before
their ServiceWorkerGlobalScope is destroyed on the running background thread.

There's no reason to not have WorkerThread guarantee its own lifetime until its background thread
has actually completed.

* workers/WorkerThread.cpp:
(WebCore::WorkerThread::workerThread): Protect the WorkerThread object during the entire runtime
  of the background thread itself, and release that protection on the main thread.
* workers/WorkerThread.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225343 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[DFG][FTL] operationHasIndexedProperty does not consider negative int32_t
utatane.tea@gmail.com [Thu, 30 Nov 2017 20:48:53 +0000 (20:48 +0000)]
[DFG][FTL] operationHasIndexedProperty does not consider negative int32_t
https://bugs.webkit.org/show_bug.cgi?id=180190

Reviewed by Mark Lam.

JSTests:

* stress/operation-in-may-have-negative-int32-array-storage.js: Added.
(shouldBe):
(test1):
* stress/operation-in-may-have-negative-int32-contiguous-array.js: Added.
(shouldBe):
(test1):
* stress/operation-in-may-have-negative-int32-double-array.js: Added.
(shouldBe):
(test1):
* stress/operation-in-may-have-negative-int32-generic-array.js: Added.
(shouldBe):
(test1):
* stress/operation-in-may-have-negative-int32-int32-array.js: Added.
(shouldBe):
(test1):
* stress/operation-in-may-have-negative-int32.js: Added.
(shouldBe):
(test2):
* stress/operation-in-negative-int32-cast.js: Added.
(shouldBe):
(test1):

Source/JavaScriptCore:

If DFG HasIndexedProperty node observes negative index, it goes to a slow
path by calling operationHasIndexedProperty. The problem is that
operationHasIndexedProperty does not account negative index. Negative index
was used as uint32 array index.

In this patch we add a path for negative index in operationHasIndexedProperty.
And rename it to operationHasIndexedPropertyByInt to make intension clear.
We also move operationHasIndexedPropertyByInt from JITOperations to DFGOperations
since it is only used in DFG and FTL.

While fixing this bug, we found that our op_in does not record OutOfBound feedback.
This causes repeated OSR exit and significantly regresses the performance. We opened
a bug to track this issue[1].

[1]: https://bugs.webkit.org/show_bug.cgi?id=180192

* dfg/DFGOperations.cpp:
* dfg/DFGOperations.h:
* dfg/DFGSpeculativeJIT32_64.cpp:
(JSC::DFG::SpeculativeJIT::compile):
* dfg/DFGSpeculativeJIT64.cpp:
(JSC::DFG::SpeculativeJIT::compile):
* ftl/FTLLowerDFGToB3.cpp:
(JSC::FTL::DFG::LowerDFGToB3::compileHasIndexedProperty):
* jit/JITOperations.cpp:
* jit/JITOperations.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225342 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoPopulate self.registration.installing/waiting/active inside service workers
cdumez@apple.com [Thu, 30 Nov 2017 20:40:04 +0000 (20:40 +0000)]
Populate self.registration.installing/waiting/active inside service workers
https://bugs.webkit.org/show_bug.cgi?id=180168

Reviewed by Brady Eidson.

LayoutTests/imported/w3c:

Rebaseline a couple of WPT that now pass for checks.

* web-platform-tests/service-workers/service-worker/ServiceWorkerGlobalScope/registration-attribute.https-expected.txt:
* web-platform-tests/service-workers/service-worker/navigation-redirect.https-expected.txt:

Source/WebCore:

Populate self.registration.installing/waiting/active inside service workers.
Also make sure the ServiceWorker's state properly gets updated inside
service workers.

No new tests, extended existing test for self.registration.

* workers/service/ServiceWorker.cpp:
(WebCore::ServiceWorker::scheduleTaskToUpdateState):
* workers/service/context/SWContextManager.cpp:
(WebCore::SWContextManager::forEachServiceWorkerThread):
* workers/service/context/SWContextManager.h:
* workers/service/context/ServiceWorkerThread.cpp:
(WebCore::ServiceWorkerThread::fireInstallEvent):
(WebCore::ServiceWorkerThread::fireActivateEvent):
* workers/service/server/SWClientConnection.cpp:
(WebCore::SWClientConnection::updateRegistrationState):
(WebCore::SWClientConnection::updateWorkerState):
(WebCore::SWClientConnection::fireUpdateFoundEvent):
* workers/service/server/SWClientConnection.h:

Source/WTF:

Add CrossThreadCopier support for std::optional<>.

* wtf/CrossThreadCopier.h:

LayoutTests:

Extend layout test coverage.

* http/tests/workers/service/resources/self_registration-worker.js:
* http/tests/workers/service/self_registration-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225341 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoUpdate ANGLE
dino@apple.com [Thu, 30 Nov 2017 20:39:54 +0000 (20:39 +0000)]
Update ANGLE
https://bugs.webkit.org/show_bug.cgi?id=180177
<rdar://problem/35774734>

Reviewed by Alex Christensen.

Huge list of changed files omitted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225340 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoUnreviewed, unskip a few Service Workers tests that no longer seem to be flaky.
cdumez@apple.com [Thu, 30 Nov 2017 19:52:19 +0000 (19:52 +0000)]
Unreviewed, unskip a few Service Workers tests that no longer seem to be flaky.

* TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225338 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoAdd a script for automatically running the buildbot configurations on a test environment.
clopez@igalia.com [Thu, 30 Nov 2017 19:29:54 +0000 (19:29 +0000)]
Add a script for automatically running the buildbot configurations on a test environment.
https://bugs.webkit.org/show_bug.cgi?id=169084

Reviewed by Michael Catanzaro.

This new script does the following:
- Creates a temporal workingdir.
- Installs buildbot and its dependencies there (very same version used by the webkit.org buildbot master https://build.webkit.org/about ).
  This is important as there are problems that that may not be detected if you don't test with the same version that will be used on production.
- Copies the buildmaster config files to the temporal workingdir.
- Performs the required steps to start a buildmaster.
- Creates the required config for each slave and starts all of them.

The test buildmaster should be accessible at http://localhost:8710/ allowing to trigger manual builds.
This allows to test any configuration changes on this test environment that is pretty much identical to the production one.

* BuildSlaveSupport/build.webkit.org-config/run-buildbot-test.py: Added.
(check_tcp_port_open):
(upgrade_db_needed):
(create_tempdir):
(print_if_error_stdout_stderr):
(setup_master_workdir):
(wait_for_master_ready):
(start_master):
(get_list_workers):
(start_worker):
(clean):
(cmd_exists):
(check_buildbot_installed):
(setup_virtualenv):
(configdir_is_valid):
(main):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225336 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoRename RegistrationOptions to ServiceWorkerRegistrationOptions
cdumez@apple.com [Thu, 30 Nov 2017 19:17:55 +0000 (19:17 +0000)]
Rename RegistrationOptions to ServiceWorkerRegistrationOptions
https://bugs.webkit.org/show_bug.cgi?id=180207

Reviewed by Geoffrey Garen.

Rename RegistrationOptions to ServiceWorkerRegistrationOptions as the name
RegistrationOptions is too generic and likely to conflict. Also modernize
ServiceWorkerRegistrationOptions' IPC decoder.

* workers/service/ServiceWorkerContainer.h:
* workers/service/ServiceWorkerJobData.h:
(WebCore::ServiceWorkerJobData::decode):
* workers/service/ServiceWorkerRegistrationOptions.cpp:
(WebCore::ServiceWorkerRegistrationOptions::isolatedCopy const):
* workers/service/ServiceWorkerRegistrationOptions.h:
(WebCore::ServiceWorkerRegistrationOptions::encode const):
(WebCore::ServiceWorkerRegistrationOptions::decode):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225335 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[build.webkit.org] Move python code from master.cfg in separate files
aakash_jain@apple.com [Thu, 30 Nov 2017 19:14:04 +0000 (19:14 +0000)]
[build.webkit.org] Move python code from cfg in separate files
https://bugs.webkit.org/show_bug.cgi?id=180180

Reviewed by Alexey Proskuryakov.

* BuildSlaveSupport/build.webkit.org-config/master.cfg: Moved code to steps.py and master.cfg appropriately.
* BuildSlaveSupport/build.webkit.org-config/factories.py: Moved all factory related code from master.cfg.
* BuildSlaveSupport/build.webkit.org-config/steps.py: Moved all steps related code from master.cfg

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225334 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoAllow JSC command line tool to accept UTF8
msaboff@apple.com [Thu, 30 Nov 2017 19:12:01 +0000 (19:12 +0000)]
Allow JSC command line tool to accept UTF8
https://bugs.webkit.org/show_bug.cgi?id=180205

Reviewed by Keith Miller.

This unifies the UTF8 handling of interactive mode with that of source files.

* jsc.cpp:
(runInteractive):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225333 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWKAttachmentTests.InPlaceImageAttachmentToggleDisplayMode times out on macOS bots
wenson_hsieh@apple.com [Thu, 30 Nov 2017 18:58:06 +0000 (18:58 +0000)]
WKAttachmentTests.InPlaceImageAttachmentToggleDisplayMode times out on macOS bots
https://bugs.webkit.org/show_bug.cgi?id=180200

Reviewed by Alexey Proskuryakov.

Rewrites an API test to avoid checking against the hard-coded platform-dependent size of an attachment element.
Instead, first insert the attachment element as an icon and compute its size, and then check that the size of
the element is restored to this original value after toggling the display mode to in-place mode and back.

* TestWebKitAPI/Tests/WebKitCocoa/WKAttachmentTests.mm:
(-[TestWKWebView attachmentElementSize]):
(-[TestWKWebView waitForAttachmentElementSizeToBecome:]):
(TestWebKitAPI::TEST):
(platformAttachmentIconElementSize): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225332 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[Mac] remove unneeded RetainPtr use introduced in r225142
darin@apple.com [Thu, 30 Nov 2017 18:17:07 +0000 (18:17 +0000)]
[Mac] remove unneeded RetainPtr use introduced in r225142
https://bugs.webkit.org/show_bug.cgi?id=180202

Reviewed by Tim Horton.

Source/WebCore:

* editing/mac/DictionaryLookup.h: Changed out arguments back to NSDictionary ** instead
of RetainPtr because these use autorelease.

* editing/mac/DictionaryLookup.mm:
(WebCore::tokenRange): Use NSDictionary **.
(WebCore::DictionaryLookup::rangeForSelection): Ditto.
(WebCore::DictionaryLookup::rangeAtHitTestResult): Ditto.
(WebCore::DictionaryLookup::stringForPDFSelection): Ditto.

Source/WebKit:

* WebProcess/Plugins/PDF/PDFPlugin.h: Got rid of unnecessary RetainPtr.
* WebProcess/Plugins/PDF/PDFPlugin.mm:
(WebKit::PDFPlugin::lookupTextAtLocation const): Ditto.
* WebProcess/WebPage/WebPage.h: Ditto.
* WebProcess/WebPage/mac/WebPageMac.mm:
(WebKit::WebPage::performDictionaryLookupAtLocation): Ditto.
(WebKit::WebPage::performDictionaryLookupForSelection): Ditto.
(WebKit::WebPage::performImmediateActionHitTestAtLocation): Ditto.
(WebKit::WebPage::lookupTextAtLocation): Ditto.

Source/WebKitLegacy/mac:

* WebView/WebImmediateActionController.mm:
(-[WebImmediateActionController _animationControllerForText]): Got rid of unnecessary
RetainPtr.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225331 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoMark ServiceWorkerGlobalScope as final
cdumez@apple.com [Thu, 30 Nov 2017 18:02:48 +0000 (18:02 +0000)]
Mark ServiceWorkerGlobalScope as final
https://bugs.webkit.org/show_bug.cgi?id=180201

Reviewed by Alex Christensen.

* workers/service/ServiceWorkerGlobalScope.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225330 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[GTK] Enable all wpt touch-events tests.
Ms2ger@igalia.com [Thu, 30 Nov 2017 16:24:48 +0000 (16:24 +0000)]
[GTK] Enable all wpt touch-events tests.
https://bugs.webkit.org/show_bug.cgi?id=180198

Unreviewed test gardening.

LayoutTests/imported/w3c:

* web-platform-tests/touch-events/create-touch-touchlist-expected.txt: Added.
* web-platform-tests/touch-events/historical-expected.txt: Renamed from LayoutTests/platform/gtk/imported/w3c/web-platform-tests/touch-events/historical-expected.txt.
* web-platform-tests/touch-events/touch-globaleventhandler-interface-expected.txt: Renamed from LayoutTests/platform/gtk/imported/w3c/web-platform-tests/touch-events/touch-globaleventhandler-interface-expected.txt.
* web-platform-tests/touch-events/touch-retargeting-expected.txt:
* web-platform-tests/touch-events/touch-touchevent-constructor-expected.txt:

LayoutTests:

* platform/gtk/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225329 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[CoordGraphics] Move CoordinatedGraphicsLayer painting behind Nicosia::PaintingEngine
zandobersek@gmail.com [Thu, 30 Nov 2017 15:12:58 +0000 (15:12 +0000)]
[CoordGraphics] Move CoordinatedGraphicsLayer painting behind Nicosia::PaintingEngine
https://bugs.webkit.org/show_bug.cgi?id=180141

Reviewed by Carlos Garcia Campos.

Source/WebCore:

Introduce a painting engine concept that's used to perform painting
for a given GraphicsLayer object.

The Nicosia::PaintingEngine object is leveraged in the
CoordinatedGraphicsLayer::updateContentBuffers() method. It's retrieved
through the CoordinatedGraphicsLayerClient interface, with a new method
introduced there for this purpose.

The Nicosia::PaintingEngine interface is kept simple for now, with only
a single paint() method. The only current implementation is the
PaintingEngineBasic class, which simply reuses the code that was
previously in CoordinatedGraphicsLayer::updateContentBuffers(),
maintaining the current functionality.

The Nicosia::PaintingEngine::create() function returns a new
PaintingEngine object. For now it defaults to PaintingEngineBasic, but
it could be configured at configure-time or even runtime in the future
to return a more advanced engine implementation.

No new tests -- no change in behavior.

* platform/TextureMapper.cmake:
* platform/graphics/nicosia/NicosiaPaintingEngine.cpp: Added.
(Nicosia::PaintingEngine::create):
* platform/graphics/nicosia/NicosiaPaintingEngine.h: Added.
* platform/graphics/nicosia/NicosiaPaintingEngineBasic.cpp: Added.
(Nicosia::PaintingEngineBasic::paint):
* platform/graphics/nicosia/NicosiaPaintingEngineBasic.h: Added.
* platform/graphics/texmap/coordinated/CoordinatedGraphicsLayer.cpp:
(WebCore::CoordinatedGraphicsLayer::updateContentBuffers):
* platform/graphics/texmap/coordinated/CoordinatedGraphicsLayer.h:

Source/WebKit:

CompositingCoordinator must implement the paintingEngine() method now
that the method's been added to the CoordinatedGraphicsLayerClient
interface. The CompositingCoordinator manages the Nicosia::PaintingEngine
object through the new m_paintingEngine member variable, and returns the
reference to this object in the paintingEngine() method.

* WebProcess/WebPage/CoordinatedGraphics/CompositingCoordinator.cpp:
(WebKit::CompositingCoordinator::CompositingCoordinator):
(WebKit::CompositingCoordinator::paintingEngine):
* WebProcess/WebPage/CoordinatedGraphics/CompositingCoordinator.h:
Drop unnecessary typedefs.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225328 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[WPE] Stop building Wayland in JHBuild moduleset
mcatanzaro@igalia.com [Thu, 30 Nov 2017 15:01:53 +0000 (15:01 +0000)]
[WPE] Stop building Wayland in JHBuild moduleset
https://bugs.webkit.org/show_bug.cgi?id=179951

Reviewed by Žan Doberšek.

Since Wayland is not used for running tests, and since it does not seem to depend on
anything in the JHBuild moduleset, I think we can remove it safely without having to worry
that the system version of Wayland might use libraries in the JHBuild environment (which
would be unsafe).

* wpe/install-dependencies: Install Wayland.
* wpe/jhbuild.modules: Stop building Wayland.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225327 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWebDriver: locator strategy should be validated before trying to find elements
carlosgc@webkit.org [Thu, 30 Nov 2017 13:48:46 +0000 (13:48 +0000)]
WebDriver: locator strategy should be validated before trying to find elements
https://bugs.webkit.org/show_bug.cgi?id=180187

Reviewed by Carlos Alberto Lopez Perez.

We currently rely on the js atom to raise an exception in case the locator strategy is not valid, but in case of
find element from element, if the element doesn't exist we fail with stale element error instead of invalid
argument as expected. So, let's validate the strategies when parsing them, which would also avoid going to the
browser in cae of invalid strategy.

Fixes: imported/w3c/webdriver/tests/retrieval/find_element_from_element.py::test_invalid_using_argument[a]
* WebDriverService.cpp:
(WebDriver::isValidStrategy):
(WebDriver::findStrategyAndSelectorOrCompleteWithError):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225326 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWebDriver: remove elementSubmit command
carlosgc@webkit.org [Thu, 30 Nov 2017 13:47:19 +0000 (13:47 +0000)]
WebDriver: remove elementSubmit command
https://bugs.webkit.org/show_bug.cgi?id=180186

Reviewed by Carlos Alberto Lopez Perez.

It's not in the spec, we had it only because selenium used it, but now it uses execute_script with custom code
to implement submit, so we can just remove it.

* Session.cpp:
(WebDriver::Session::elementSubmit): Deleted.
* Session.h:
* WebDriverService.cpp:
(WebDriver::WebDriverService::elementSubmit): Deleted.
* WebDriverService.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225325 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoAdd StyleBench
antti@apple.com [Thu, 30 Nov 2017 12:30:25 +0000 (12:30 +0000)]
Add StyleBench
https://bugs.webkit.org/show_bug.cgi?id=180140

Reviewed by Simon Fraser and Joseph Pecoraro.

StyleBench tests performance of the CSS style resolution and style invalidation. Each test run
creates a large document and a large stylesheet using varying settings. It then applies
a series of mutations to the document and measures the time to update the style and rendering.
The resulting layout is simple, most of the pressure is on selector matching.

StyleBench uses Speedometer framework for UI and measurements. For profiling purposes, it can also
be run locally by opening style-bench.html directly.

There are currently four subtests:

- child and descendant combinators only (all other tests have these too).
- sibling combinators: '~' and '+'
- positional pseudo classes: :nth-child and similar
- ::before and ::after pseudo elements

The measured DOM mutations are:

- add classes
- remove classes
- add leaf elements
- remove leaf elements

* StyleBench: Added.
* StyleBench/InteractiveRunner.html: Added.

    Copied and customized from Speedometer.

* StyleBench/index.html: Added.

    Copied and customized from Speedometer.

* StyleBench/resources: Added.
* StyleBench/resources/style-bench.html: Added.
* StyleBench/resources/style-bench.js: Added.

    The test class.

(Random):
(Random.prototype.get next):
(Random.prototype.chance):
(Random.prototype.number):
(nextAnimationFrame):
(defaultConfiguration):
(descendantCombinatorConfiguration):
(siblingCombinatorConfiguration):
(pseudoClassConfiguration):
(beforeAndAfterConfiguration):
(predefinedConfigurations):

    Four predefined configurations.

(prototype.randomElementName):
(prototype.randomCombinator):
(prototype.randomPseudoClass):
(prototype.makeSimpleSelector):
(prototype.makeSelector):
(prototype.get randomColorComponent):
(prototype.makeDeclaration):
(prototype.makeRule):
(prototype.makeStylesheet):
(prototype.makeStyle):
(prototype.makeElement):
(prototype.makeTreeWithDepth):
(prototype.makeTree):
(prototype.updateCachedTestElements):
(prototype.randomTreeElement):
(prototype.addClasses):
(prototype.removeClasses):
(prototype.addLeafElements):
(prototype.removeLeafElements):
(prototype.async.runForever):
* StyleBench/resources/tests.js: Added.
(makeSteps):
(makeSuite):

    Generates Speedometer Suites.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225324 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[GStreamer] Builds fails with ENABLE_VIDEO=OFF due to GStreamer usage
aperez@igalia.com [Thu, 30 Nov 2017 12:00:56 +0000 (12:00 +0000)]
[GStreamer] Builds fails with ENABLE_VIDEO=OFF due to GStreamer usage
https://bugs.webkit.org/show_bug.cgi?id=179373

Reviewed by Xabier Rodriguez-Calvar.

This allows to build with WebAudio enabled even if video support is
disabled by guarding usage of HTMLMediaElement with ENABLE(VIDEO).

No new tests needed.

* page/DeprecatedGlobalSettings.cpp:
(WebCore::DeprecatedGlobalSettings::setGStreamerEnabled): Guard usage of HTMLMediaElement.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225323 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoREGRESSION(r225314): [Linux] More than 2000 jsc tests are failing after r225314
utatane.tea@gmail.com [Thu, 30 Nov 2017 11:27:43 +0000 (11:27 +0000)]
REGRESSION(r225314): [Linux] More than 2000 jsc tests are failing after r225314
https://bugs.webkit.org/show_bug.cgi?id=180185

Reviewed by Carlos Garcia Campos.

After r225314, we start using AllocatorForMode::MustAlreadyHaveAllocator for JSRopeString's allocatorFor.
But it is different from the original code used before r225314. Since DFGSpeculativeJIT::emitAllocateJSCell
can accept nullptr allocator, the behavior of the original code is AllocatorForMode::AllocatorIfExists.
And JSRopeString's allocator may not exist at this point if any JSRopeString is not allocated. But MakeRope
DFG node can be emitted if we see untaken path includes String + String code.

This patch fixes Linux JSC crashes by changing JSRopeString's AllocatorForMode to AllocatorIfExists.
As a result, only one user of AllocatorForMode::MustAlreadyHaveAllocator is MaterializeNewObject in FTL.
I'm not sure why this condition (MustAlreadyHaveAllocator) is ensured. But this code is the same to the
original code used before r225314.

* dfg/DFGSpeculativeJIT.cpp:
(JSC::DFG::SpeculativeJIT::compileMakeRope):
* ftl/FTLLowerDFGToB3.cpp:
(JSC::FTL::DFG::LowerDFGToB3::compileMakeRope):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225322 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoSilence the BitVector leak for
ap@apple.com [Thu, 30 Nov 2017 08:21:25 +0000 (08:21 +0000)]
Silence the BitVector leak for
https://bugs.webkit.org/show_bug.cgi?id=121662

The out of line data pointer isn't stored directly, so the leaks tool cannot see it.

* Scripts/webkitpy/port/leakdetector.py:
(LeakDetector._callstacks_to_exclude_from_leaks):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225321 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWTF / bmalloc: don't write to 0xbbadbeef when ASAN is looking
jfbastien@apple.com [Thu, 30 Nov 2017 07:53:21 +0000 (07:53 +0000)]
WTF / bmalloc: don't write to 0xbbadbeef when ASAN is looking
https://bugs.webkit.org/show_bug.cgi?id=180175

Reviewed by Mark Lam.

ASAN knows that 0xbbadbeef is a bbad aaddress, and tells us so
when we write to it, say in an assert. That creates bbad error
reports where ASAN thinks we write to an invalid address, instead
of thinking that we hit an assertion. In some cases, tooling that
use fuzzers aggregate similar issues, and think that we just have
the one bug and not a bunch of different asserts.

Source/bmalloc:

At the same time, bmalloc's version of CRASH just writes to
0xbbadbeef and assumes that's invalid and will crash, which isn't
necessarily true on non-Mac platforms. WTF's version then makes
sure there's a crash, so bmalloc should do the same.

* bmalloc.xcodeproj/project.pbxproj:
* bmalloc/BAssert.h:
* bmalloc/BCompiler.h: Added.
* bmalloc/BPlatform.h:

Source/WTF:

* wtf/Assertions.cpp:
* wtf/Assertions.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225320 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoFix build.webkit.org broken unit-test after r225080
aakash_jain@apple.com [Thu, 30 Nov 2017 07:45:55 +0000 (07:45 +0000)]
Fix build.webkit.org broken unit-test after r225080
https://bugs.webkit.org/show_bug.cgi?id=180176

Reviewed by Carlos Garcia Campos.

* BuildSlaveSupport/build.webkit.org-config/mastercfg_unittest.py: Updated expected_build_steps.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225319 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoMove modern media controls test expectations from ios-simulator to ios
graouts@webkit.org [Thu, 30 Nov 2017 06:15:24 +0000 (06:15 +0000)]
Move modern media controls test expectations from ios-simulator to ios
https://bugs.webkit.org/show_bug.cgi?id=180158

Reviewed by Dean Jackson.

We shouldn't be using the ios-simulator platform for these, ios is the right one.

* platform/ios-simulator/TestExpectations:
* platform/ios/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225317 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoServiceWorker WebProcess sometimes crashes in JSVMClientData::~JSVMClientData()
cdumez@apple.com [Thu, 30 Nov 2017 06:13:40 +0000 (06:13 +0000)]
ServiceWorker WebProcess sometimes crashes in JSVMClientData::~JSVMClientData()
https://bugs.webkit.org/show_bug.cgi?id=180173

Reviewed by Alex Christensen.

The leak was caused by EventListeners remaining when destroying the VM, because
JSEventListener refs the DOMWrapperWorld. To address the issue, we now call
removeAllEventListeners() in the stop() method of ServiceWorkerContainer,
ServiceWorkerRegistration and ServiceWorker. Those event listeners are no
longer needed after ActiveDOMObject::stop() is called since the script
execution context is about to be destroyed.

This is the same pattern used in IDBDatabase::stop(), IDBRequest::stop().

No new tests, already covered by existing test.

* workers/service/ServiceWorker.cpp:
(WebCore::ServiceWorker::stop):
* workers/service/ServiceWorkerContainer.cpp:
(WebCore::ServiceWorkerContainer::stop):
* workers/service/ServiceWorkerContainer.h:
* workers/service/ServiceWorkerRegistration.cpp:
(WebCore::ServiceWorkerRegistration::stop):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225316 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoCodeBlockSet::deleteUnmarkedAndUnreferenced can be a little more efficient
fpizlo@apple.com [Thu, 30 Nov 2017 04:48:52 +0000 (04:48 +0000)]
CodeBlockSet::deleteUnmarkedAndUnreferenced can be a little more efficient
https://bugs.webkit.org/show_bug.cgi?id=180108

Reviewed by Saam Barati.

This was creating a vector of things to remove and then removing them. I think I remember writing
this code, and I did that because at the time we did not have removeAllMatching, which is
definitely better. This is a minuscule optimization for Speedometer. I wanted to land this
obvious improvement before I did more fundamental things to this code.

* heap/CodeBlockSet.cpp:
(JSC::CodeBlockSet::deleteUnmarkedAndUnreferenced):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225315 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoGC should support isoheaps
fpizlo@apple.com [Thu, 30 Nov 2017 04:39:50 +0000 (04:39 +0000)]
GC should support isoheaps
https://bugs.webkit.org/show_bug.cgi?id=179288

Reviewed by Saam Barati.
Source/JavaScriptCore:

This expands the power of the Subspace API in JSC:

- Everything associated with describing the types of objects is now part of the HeapCellType class.
  We have different HeapCellTypes for different destruction strategies. Any Subspace can use any
  HeapCellType; these are orthogonal things.

- There are now two variants of Subspace: CompleteSubspace, which can allocate any size objects using
  any AlignedMemoryAllocator; and IsoSubspace, which can allocate just one size of object and uses a
  special virtual memory pool for that purpose. Like bmalloc's IsoHeap, IsoSubspace hoards virtual
  pages but releases the physical pages as part of the respective allocator's scavenging policy
  (the Scavenger in bmalloc for IsoHeap and the incremental sweep and full sweep in Riptide for
  IsoSubspace).

So far, this patch just puts subtypes of ExecutableBase in IsoSubspaces. If it works, we can use it
for more things.

This does not have any effect on JetStream (0.18% faster with p = 0.69).

* JavaScriptCore.xcodeproj/project.pbxproj:
* Sources.txt:
* bytecode/AccessCase.cpp:
(JSC::AccessCase::generateImpl):
* bytecode/ObjectAllocationProfileInlines.h:
(JSC::ObjectAllocationProfile::initializeProfile):
* dfg/DFGSpeculativeJIT.cpp:
(JSC::DFG::SpeculativeJIT::emitAllocateRawObject):
(JSC::DFG::SpeculativeJIT::compileMakeRope):
(JSC::DFG::SpeculativeJIT::compileAllocatePropertyStorage):
(JSC::DFG::SpeculativeJIT::compileReallocatePropertyStorage):
* dfg/DFGSpeculativeJIT64.cpp:
(JSC::DFG::SpeculativeJIT::compile):
* ftl/FTLAbstractHeapRepository.h:
* ftl/FTLLowerDFGToB3.cpp:
(JSC::FTL::DFG::LowerDFGToB3::compileMakeRope):
(JSC::FTL::DFG::LowerDFGToB3::compileMaterializeNewObject):
(JSC::FTL::DFG::LowerDFGToB3::allocatePropertyStorageWithSizeImpl):
(JSC::FTL::DFG::LowerDFGToB3::allocateObject):
(JSC::FTL::DFG::LowerDFGToB3::allocatorForSize):
* heap/AlignedMemoryAllocator.cpp:
(JSC::AlignedMemoryAllocator::registerAllocator):
(JSC::AlignedMemoryAllocator::registerSubspace):
* heap/AlignedMemoryAllocator.h:
(JSC::AlignedMemoryAllocator::firstAllocator const):
* heap/AllocationFailureMode.h: Added.
* heap/CompleteSubspace.cpp: Added.
(JSC::CompleteSubspace::CompleteSubspace):
(JSC::CompleteSubspace::~CompleteSubspace):
(JSC::CompleteSubspace::allocatorFor):
(JSC::CompleteSubspace::allocate):
(JSC::CompleteSubspace::allocateNonVirtual):
(JSC::CompleteSubspace::allocatorForSlow):
(JSC::CompleteSubspace::allocateSlow):
(JSC::CompleteSubspace::tryAllocateSlow):
* heap/CompleteSubspace.h: Added.
(JSC::CompleteSubspace::offsetOfAllocatorForSizeStep):
(JSC::CompleteSubspace::allocatorForSizeStep):
(JSC::CompleteSubspace::allocatorForNonVirtual):
* heap/HeapCellType.cpp: Added.
(JSC::HeapCellType::HeapCellType):
(JSC::HeapCellType::~HeapCellType):
(JSC::HeapCellType::finishSweep):
(JSC::HeapCellType::destroy):
* heap/HeapCellType.h: Added.
(JSC::HeapCellType::attributes const):
* heap/IsoAlignedMemoryAllocator.cpp: Added.
(JSC::IsoAlignedMemoryAllocator::IsoAlignedMemoryAllocator):
(JSC::IsoAlignedMemoryAllocator::~IsoAlignedMemoryAllocator):
(JSC::IsoAlignedMemoryAllocator::tryAllocateAlignedMemory):
(JSC::IsoAlignedMemoryAllocator::freeAlignedMemory):
(JSC::IsoAlignedMemoryAllocator::dump const):
* heap/IsoAlignedMemoryAllocator.h: Added.
* heap/IsoSubspace.cpp: Added.
(JSC::IsoSubspace::IsoSubspace):
(JSC::IsoSubspace::~IsoSubspace):
(JSC::IsoSubspace::allocatorFor):
(JSC::IsoSubspace::allocatorForNonVirtual):
(JSC::IsoSubspace::allocate):
(JSC::IsoSubspace::allocateNonVirtual):
* heap/IsoSubspace.h: Added.
(JSC::IsoSubspace::size const):
* heap/MarkedAllocator.cpp:
(JSC::MarkedAllocator::MarkedAllocator):
(JSC::MarkedAllocator::setSubspace):
(JSC::MarkedAllocator::allocateSlowCase):
(JSC::MarkedAllocator::tryAllocateSlowCase): Deleted.
(JSC::MarkedAllocator::allocateSlowCaseImpl): Deleted.
* heap/MarkedAllocator.h:
(JSC::MarkedAllocator::nextAllocatorInAlignedMemoryAllocator const):
(JSC::MarkedAllocator::setNextAllocatorInAlignedMemoryAllocator):
* heap/MarkedAllocatorInlines.h:
(JSC::MarkedAllocator::allocate):
(JSC::MarkedAllocator::tryAllocate): Deleted.
* heap/MarkedBlock.h:
* heap/MarkedBlockInlines.h:
(JSC::MarkedBlock::Handle::finishSweepKnowingHeapCellType):
(JSC::MarkedBlock::Handle::finishSweepKnowingSubspace): Deleted.
* heap/MarkedSpace.cpp:
(JSC::MarkedSpace::addMarkedAllocator):
* heap/MarkedSpace.h:
* heap/Subspace.cpp:
(JSC::Subspace::Subspace):
(JSC::Subspace::initialize):
(JSC::Subspace::finishSweep):
(JSC::Subspace::destroy):
(JSC::Subspace::prepareForAllocation):
(JSC::Subspace::findEmptyBlockToSteal):
(): Deleted.
(JSC::Subspace::allocate): Deleted.
(JSC::Subspace::tryAllocate): Deleted.
(JSC::Subspace::allocatorForSlow): Deleted.
(JSC::Subspace::allocateSlow): Deleted.
(JSC::Subspace::tryAllocateSlow): Deleted.
(JSC::Subspace::didAllocate): Deleted.
* heap/Subspace.h:
(JSC::Subspace::heapCellType const):
(JSC::Subspace::nextSubspaceInAlignedMemoryAllocator const):
(JSC::Subspace::setNextSubspaceInAlignedMemoryAllocator):
(JSC::Subspace::offsetOfAllocatorForSizeStep): Deleted.
(JSC::Subspace::allocatorForSizeStep): Deleted.
(JSC::Subspace::tryAllocatorFor): Deleted.
(JSC::Subspace::allocatorFor): Deleted.
* jit/AssemblyHelpers.h:
(JSC::AssemblyHelpers::emitAllocateJSObjectWithKnownSize):
(JSC::AssemblyHelpers::emitAllocateVariableSized):
(JSC::AssemblyHelpers::emitAllocateVariableSizedCell):
* jit/JITOpcodes.cpp:
(JSC::JIT::emit_op_new_object):
* runtime/ButterflyInlines.h:
(JSC::Butterfly::createUninitialized):
(JSC::Butterfly::tryCreate):
(JSC::Butterfly::growArrayRight):
* runtime/DirectArguments.cpp:
(JSC::DirectArguments::overrideThings):
* runtime/DirectArguments.h:
(JSC::DirectArguments::subspaceFor):
* runtime/DirectEvalExecutable.h:
* runtime/EvalExecutable.h:
* runtime/ExecutableBase.h:
(JSC::ExecutableBase::subspaceFor):
* runtime/FunctionExecutable.h:
* runtime/GenericArgumentsInlines.h:
(JSC::GenericArguments<Type>::initModifiedArgumentsDescriptor):
* runtime/HashMapImpl.h:
(JSC::HashMapBuffer::create):
* runtime/IndirectEvalExecutable.h:
* runtime/JSArray.cpp:
(JSC::JSArray::tryCreateUninitializedRestricted):
(JSC::JSArray::unshiftCountSlowCase):
* runtime/JSArray.h:
(JSC::JSArray::tryCreate):
* runtime/JSArrayBufferView.cpp:
(JSC::JSArrayBufferView::ConstructionContext::ConstructionContext):
* runtime/JSCell.h:
(JSC::subspaceFor):
* runtime/JSCellInlines.h:
(JSC::JSCell::subspaceFor):
(JSC::tryAllocateCellHelper):
(JSC::allocateCell):
(JSC::tryAllocateCell):
* runtime/JSDestructibleObject.h:
(JSC::JSDestructibleObject::subspaceFor):
* runtime/JSDestructibleObjectHeapCellType.cpp: Copied from Source/JavaScriptCore/runtime/JSDestructibleObjectSubspace.cpp.
(JSC::JSDestructibleObjectHeapCellType::JSDestructibleObjectHeapCellType):
(JSC::JSDestructibleObjectHeapCellType::~JSDestructibleObjectHeapCellType):
(JSC::JSDestructibleObjectHeapCellType::finishSweep):
(JSC::JSDestructibleObjectHeapCellType::destroy):
(JSC::JSDestructibleObjectSubspace::JSDestructibleObjectSubspace): Deleted.
(JSC::JSDestructibleObjectSubspace::~JSDestructibleObjectSubspace): Deleted.
(JSC::JSDestructibleObjectSubspace::finishSweep): Deleted.
(JSC::JSDestructibleObjectSubspace::destroy): Deleted.
* runtime/JSDestructibleObjectHeapCellType.h: Copied from Source/JavaScriptCore/runtime/JSDestructibleObjectSubspace.h.
* runtime/JSDestructibleObjectSubspace.cpp: Removed.
* runtime/JSDestructibleObjectSubspace.h: Removed.
* runtime/JSLexicalEnvironment.h:
(JSC::JSLexicalEnvironment::subspaceFor):
* runtime/JSSegmentedVariableObject.h:
(JSC::JSSegmentedVariableObject::subspaceFor):
* runtime/JSSegmentedVariableObjectHeapCellType.cpp: Copied from Source/JavaScriptCore/runtime/JSSegmentedVariableObjectSubspace.cpp.
(JSC::JSSegmentedVariableObjectHeapCellType::JSSegmentedVariableObjectHeapCellType):
(JSC::JSSegmentedVariableObjectHeapCellType::~JSSegmentedVariableObjectHeapCellType):
(JSC::JSSegmentedVariableObjectHeapCellType::finishSweep):
(JSC::JSSegmentedVariableObjectHeapCellType::destroy):
(JSC::JSSegmentedVariableObjectSubspace::JSSegmentedVariableObjectSubspace): Deleted.
(JSC::JSSegmentedVariableObjectSubspace::~JSSegmentedVariableObjectSubspace): Deleted.
(JSC::JSSegmentedVariableObjectSubspace::finishSweep): Deleted.
(JSC::JSSegmentedVariableObjectSubspace::destroy): Deleted.
* runtime/JSSegmentedVariableObjectHeapCellType.h: Copied from Source/JavaScriptCore/runtime/JSSegmentedVariableObjectSubspace.h.
* runtime/JSSegmentedVariableObjectSubspace.cpp: Removed.
* runtime/JSSegmentedVariableObjectSubspace.h: Removed.
* runtime/JSString.h:
(JSC::JSString::subspaceFor):
* runtime/JSStringHeapCellType.cpp: Copied from Source/JavaScriptCore/runtime/JSStringSubspace.cpp.
(JSC::JSStringHeapCellType::JSStringHeapCellType):
(JSC::JSStringHeapCellType::~JSStringHeapCellType):
(JSC::JSStringHeapCellType::finishSweep):
(JSC::JSStringHeapCellType::destroy):
(JSC::JSStringSubspace::JSStringSubspace): Deleted.
(JSC::JSStringSubspace::~JSStringSubspace): Deleted.
(JSC::JSStringSubspace::finishSweep): Deleted.
(JSC::JSStringSubspace::destroy): Deleted.
* runtime/JSStringHeapCellType.h: Copied from Source/JavaScriptCore/runtime/JSStringSubspace.h.
* runtime/JSStringSubspace.cpp: Removed.
* runtime/JSStringSubspace.h: Removed.
* runtime/ModuleProgramExecutable.h:
* runtime/NativeExecutable.h:
* runtime/ProgramExecutable.h:
* runtime/RegExpMatchesArray.h:
(JSC::tryCreateUninitializedRegExpMatchesArray):
* runtime/ScopedArguments.h:
(JSC::ScopedArguments::subspaceFor):
* runtime/VM.cpp:
(JSC::VM::VM):
* runtime/VM.h:
(JSC::VM::gigacageAuxiliarySpace):
* wasm/js/JSWebAssemblyCodeBlock.h:
* wasm/js/JSWebAssemblyCodeBlockHeapCellType.cpp: Copied from Source/JavaScriptCore/wasm/js/JSWebAssemblyCodeBlockSubspace.cpp.
(JSC::JSWebAssemblyCodeBlockHeapCellType::JSWebAssemblyCodeBlockHeapCellType):
(JSC::JSWebAssemblyCodeBlockHeapCellType::~JSWebAssemblyCodeBlockHeapCellType):
(JSC::JSWebAssemblyCodeBlockHeapCellType::finishSweep):
(JSC::JSWebAssemblyCodeBlockHeapCellType::destroy):
(JSC::JSWebAssemblyCodeBlockSubspace::JSWebAssemblyCodeBlockSubspace): Deleted.
(JSC::JSWebAssemblyCodeBlockSubspace::~JSWebAssemblyCodeBlockSubspace): Deleted.
(JSC::JSWebAssemblyCodeBlockSubspace::finishSweep): Deleted.
(JSC::JSWebAssemblyCodeBlockSubspace::destroy): Deleted.
* wasm/js/JSWebAssemblyCodeBlockHeapCellType.h: Copied from Source/JavaScriptCore/wasm/js/JSWebAssemblyCodeBlockSubspace.h.
* wasm/js/JSWebAssemblyCodeBlockSubspace.cpp: Removed.
* wasm/js/JSWebAssemblyCodeBlockSubspace.h: Removed.
* wasm/js/JSWebAssemblyMemory.h:
(JSC::JSWebAssemblyMemory::subspaceFor):

Source/WebCore:

No new tests because no new behavior.

Adopting changes in JSC Subspace API.

* ForwardingHeaders/runtime/JSDestructibleObjectHeapCellType.h: Added.
* ForwardingHeaders/runtime/JSSegmentedVariableObjectHeapCellType.h: Added.
* bindings/js/JSDOMWrapper.cpp:
(WebCore::outputConstraintSubspaceFor):
(WebCore::globalObjectOutputConstraintSubspaceFor):
* bindings/js/JSDOMWrapper.h:
* bindings/js/WebCoreJSClientData.cpp:
(WebCore::JSVMClientData::JSVMClientData):
* bindings/js/WebCoreJSClientData.h:
(WebCore::JSVMClientData::outputConstraintSpace):
(WebCore::JSVMClientData::globalObjectOutputConstraintSpace):
* bindings/scripts/CodeGeneratorJS.pm:
(GenerateHeader):

Source/WTF:

One of my favorite data structures in the GC is a singly-linked list that knows its tail, so that
things get added to it at the end rather that at the beginning. In this patch, I use this to put
the same node on multiple lists, which our existing linked list templates also don't support.

This adds a new linked list that does those things:

- It supports append(). It could also support prepend(), but currently there is no need for it.

- It supports nodes that are on multiple lists. The GC uses std::mem_fn() to create a lambda that the
  list uses to set next.

* WTF.xcodeproj/project.pbxproj:
* wtf/SinglyLinkedListWithTail.h: Added.
(WTF::SinglyLinkedListWithTail::isEmpty const):
(WTF::SinglyLinkedListWithTail::append):
(WTF::SinglyLinkedListWithTail::first const):
(WTF::SinglyLinkedListWithTail::last const):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225314 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[MediaStream] Clean up audio and video capture factories
eric.carlson@apple.com [Thu, 30 Nov 2017 03:36:26 +0000 (03:36 +0000)]
[MediaStream] Clean up audio and video capture factories
https://bugs.webkit.org/show_bug.cgi?id=180156
<rdar://problem/35753435>

Reviewed by Youenn Fablet.

No new tests, no behavior change.

* platform/mediastream/RealtimeMediaSourceCenter.cpp:
(WebCore::RealtimeMediaSourceCenter::setAudioFactory): Deleted.
(WebCore::RealtimeMediaSourceCenter::unsetAudioFactory): Deleted.
(WebCore::RealtimeMediaSourceCenter::audioFactory): Deleted.
(WebCore::RealtimeMediaSourceCenter::setVideoFactory): Deleted.
(WebCore::RealtimeMediaSourceCenter::unsetVideoFactory): Deleted.
(WebCore::RealtimeMediaSourceCenter::videoFactory): Deleted.
(WebCore::RealtimeMediaSourceCenter::setAudioCaptureDeviceManager): Deleted.
(WebCore::RealtimeMediaSourceCenter::unsetAudioCaptureDeviceManager): Deleted.
(WebCore::RealtimeMediaSourceCenter::audioCaptureDeviceManager): Deleted.
(WebCore::RealtimeMediaSourceCenter::setVideoCaptureDeviceManager): Deleted.
(WebCore::RealtimeMediaSourceCenter::unsetVideoCaptureDeviceManager): Deleted.
(WebCore::RealtimeMediaSourceCenter::videoCaptureDeviceManager): Deleted.
* platform/mediastream/RealtimeMediaSourceCenter.h:
(WebCore::RealtimeMediaSourceCenter::setAudioFactory):
(WebCore::RealtimeMediaSourceCenter::unsetAudioFactory):
* platform/mediastream/RealtimeMediaSourceSettings.cpp:
(WebCore::RealtimeMediaSourceSettings::facingMode):
(WebCore::RealtimeMediaSourceSettings::videoFacingModeEnum):
(WebCore::userFacing): Deleted.
(WebCore::environmentFacing): Deleted.
(WebCore::leftFacing): Deleted.
(WebCore::rightFacing): Deleted.
* platform/mediastream/RealtimeMediaSourceSettings.h:
(WebCore::RealtimeMediaSourceSettings::RealtimeMediaSourceSettings): Deleted.
* platform/mediastream/mac/AVVideoCaptureSource.h:
* platform/mediastream/mac/AVVideoCaptureSource.mm:
(WebCore::AVVideoCaptureSource::create):
(WebCore::AVVideoCaptureSource::~AVVideoCaptureSource):
(WebCore::AVVideoCaptureSource::setupCaptureSession):
(): Deleted.
(WebCore::AVVideoCaptureSourceFactory::setVideoCapturePageState): Deleted.
(WebCore::avVideoCaptureSourceFactory): Deleted.
(WebCore::AVVideoCaptureSource::factory): Deleted.
* platform/mediastream/mac/RealtimeMediaSourceCenterMac.cpp:
(WebCore::videoCaptureSourceFactory):
(WebCore::RealtimeMediaSourceCenterMac::audioFactory):
(WebCore::RealtimeMediaSourceCenterMac::videoFactory):
(WebCore::RealtimeMediaSourceCenterMac::audioCaptureDeviceManager):
(WebCore::RealtimeMediaSourceCenterMac::videoCaptureDeviceManager):
(WebCore::RealtimeMediaSourceCenterMac::RealtimeMediaSourceCenterMac): Deleted.
(WebCore::RealtimeMediaSourceCenterMac::defaultAudioFactory): Deleted.
(WebCore::RealtimeMediaSourceCenterMac::defaultVideoFactory): Deleted.
(WebCore::RealtimeMediaSourceCenterMac::defaultAudioCaptureDeviceManager): Deleted.
(WebCore::RealtimeMediaSourceCenterMac::defaultVideoCaptureDeviceManager): Deleted.
* platform/mediastream/mac/RealtimeMediaSourceCenterMac.h:
* platform/mock/MockRealtimeAudioSource.cpp:
(WebCore::MockRealtimeAudioSource::~MockRealtimeAudioSource):
(WebCore::MockRealtimeAudioSource::startProducingData):
* platform/mock/MockRealtimeMediaSourceCenter.h:
* platform/mock/MockRealtimeVideoSource.cpp:
(WebCore::MockRealtimeVideoSource::~MockRealtimeVideoSource):
(WebCore::MockRealtimeVideoSource::startProducingData):
* platform/mock/MockRealtimeVideoSource.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225313 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoLayoutTest http/tests/workers/service/registration-clear-redundant-worker.html is...
cdumez@apple.com [Thu, 30 Nov 2017 02:31:43 +0000 (02:31 +0000)]
LayoutTest http/tests/workers/service/registration-clear-redundant-worker.html is a flaky failure
https://bugs.webkit.org/show_bug.cgi?id=179778

Reviewed by Youenn Fablet.

As per the specification [1], when clearing a registration, the registration's workers are first
marked as redundant *before* the registration's worker is set to null.

The test was waiting for the service worker's state to become redundant and then checking in a
setTimeout(0) that the registration's worker became null. However, a setTimeout(0) is racy here.
To address the issue, we now use setInterval() and wait until the worker becomes null.

[1] https://w3c.github.io/ServiceWorker/#clear-registration-algorithm

* TestExpectations:
* http/tests/workers/service/registration-clear-redundant-worker.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225312 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[Attachment Support] Implement SPI for clients to make an attachment element display...
wenson_hsieh@apple.com [Thu, 30 Nov 2017 02:23:44 +0000 (02:23 +0000)]
[Attachment Support] Implement SPI for clients to make an attachment element display in-place
https://bugs.webkit.org/show_bug.cgi?id=180153
<rdar://problem/35735252>

Reviewed by Tim Horton.

Source/WebCore:

Adds SPI support for inserting attachments using in-place display style, and updating display options for
existing attachments. See comments below for more detail.

Tests: WKAttachmentTests.InPlaceImageAttachmentToggleDisplayMode
       WKAttachmentTests.InPlaceImageAttachmentParagraphInsertion
       WKAttachmentTests.InPlaceVideoAttachmentInsertionWithinList
       WKAttachmentTests.InPlacePDFAttachmentCutAndPaste

* WebCore.xcodeproj/project.pbxproj:
* editing/Editor.cpp:
(WebCore::Editor::insertAttachment):
(WebCore::Editor::insertAttachmentFromFile):

Update display options for the attachment before inserting into the document.

* editing/Editor.h:
* html/AttachmentTypes.h: Added.

Add a new header to define the new attachment display types. This lets us avoid importing HTMLAttachmentElement.h
and instead just import AttachmentTypes.h in some places in WebKit that only deal with plumbing
AttachmentDisplayOptions to the web process.

(WebCore::AttachmentDisplayOptions::encode const):
(WebCore::AttachmentDisplayOptions::decode):

Support serializing and deserializing attachment display options.

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

Regenerate the shadow root if needed when setting the file.

(WebCore::HTMLAttachmentElement::updateDisplayMode):

Introduce a new method to update the display mode of an attachment element. This builds up the shadow root of
the attachment if it is displayed in-place.

(WebCore::HTMLAttachmentElement::ensureInnerImage):
(WebCore::HTMLAttachmentElement::ensureInnerVideo):

Helpers to insert image and video elements into the shadow root if needed, and return the image or video element.

(WebCore::HTMLAttachmentElement::innerImage const):
(WebCore::HTMLAttachmentElement::innerVideo const):

Helpers to retrieve existing image and video elements in the shadow root.

(WebCore::HTMLAttachmentElement::populateShadowRootIfNecessary):

Appends and configures the attachment element's shadow root for inline display, initializing an image or video
element or neither, depending on the attachment type.

* html/HTMLAttachmentElement.h:

Introduce an AttachmentDisplayOptions struct which mirrors _WKAttachmentDisplayOptions in the WebKit layer. For
now, this only contains a display mode enum.

* platform/audio/ios/AudioSessionIOS.mm:

Avoid this AVAudioSession assertion when targeting iOS simulator. AVAudioSession always throws this error when
setting an audio session category in the simulator, even in a basic test app, since AVFoundation doesn't support
audio sessions in the simulator.

(WebCore::AudioSession::setCategory):

Source/WebKit:

Add new WebKit SPI, -[_WKAttachment setDisplayOptions:completion:], and add plumbing for attachment display
options to the web content process. Changes covered by 4 new API tests.

* Scripts/webkit/messages.py:
(headers_for_type):
* UIProcess/API/APIAttachment.cpp:
(API::Attachment::setDisplayOptions):
* UIProcess/API/APIAttachment.h:
* UIProcess/API/Cocoa/WKWebView.mm:
(-[WKWebView _insertAttachmentWithFilename:contentType:data:options:completion:]):

Respect given display options when inserting a new attachment.

* UIProcess/API/Cocoa/_WKAttachment.h:
* UIProcess/API/Cocoa/_WKAttachment.mm:
(-[_WKAttachmentDisplayOptions coreDisplayOptions]):

Introduce a helper to convert from the Cocoa _WKAttachmentDisplayOptions object to platform-agnostic
AttachmentDisplayOptions.

(-[_WKAttachment setDisplayOptions:completion:]):
(WebKit::if): Deleted.
* UIProcess/API/Cocoa/_WKAttachmentInternal.h:
* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::insertAttachment):
(WebKit::WebPageProxy::setAttachmentDisplayOptions):
* UIProcess/WebPageProxy.h:
* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::insertAttachment):
(WebKit::WebPage::requestAttachmentData):
(WebKit::WebPage::setAttachmentDisplayOptions):
(WebKit::WebPage::attachmentElementWithIdentifier const):

Pull common logic to retrieve an attachment element matching a given identifier out into a helper.

* WebProcess/WebPage/WebPage.h:
* WebProcess/WebPage/WebPage.messages.in:

Tools:

Adds new API tests to cover inserting in-place attachments and updating the display mode of existing attachments,
as well as performing a few editing operations (paragraph insertion, cut/paste, list insertion) on attachment
elements.

* TestWebKitAPI/Tests/WebKitCocoa/WKAttachmentTests.mm:
(platformAttachmentIconElementSize):
(testVideoData):
(testPDFData):
(displayOptionsWithMode):
(-[TestWKWebView waitForAttachmentElementSizeToBecome:]):
(-[_WKAttachment synchronouslySetDisplayOptions:error:]):
(TestWebKitAPI::TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225311 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoMissing layer content when animating elements on-screen
simon.fraser@apple.com [Thu, 30 Nov 2017 01:50:14 +0000 (01:50 +0000)]
Missing layer content when animating elements on-screen
https://bugs.webkit.org/show_bug.cgi?id=180178
rdar://problem/34923438

Reviewed by Dean Jackson.

If a delayed animation starts, that animates layers from offscreen, then we would fail
to run the logic that ensures that those layers have backing store.

Fix by ensuring that if any layer starts or ends a transform animation, we re-evaluate
backing store attachment on all its descendants.

I tried to make a test, but layer flushing is timing-sensitive and the test would have taken
5s, and not been reliable. There's a manual test in the bug.

* platform/graphics/ca/GraphicsLayerCA.cpp:
(WebCore::GraphicsLayerCA::recursiveCommitChanges):
* platform/graphics/ca/GraphicsLayerCA.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225310 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoPart 2: Adopt updated NSKeyed[Un]Archiver API when available
bfulgham@apple.com [Thu, 30 Nov 2017 01:40:56 +0000 (01:40 +0000)]
Part 2: Adopt updated NSKeyed[Un]Archiver API when available
https://bugs.webkit.org/show_bug.cgi?id=180127
<rdar://problem/35710738>

Reviewed by Simon Fraser.

The API that accepts a user-allocated NSMutableData is deprecated. Switch (for macOS 10.12 and newer)
to the modern API. Use the original API for macOS builds prior to 10.12.

Source/WebCore/PAL:

* pal/spi/cocoa/NSKeyedArchiverSPI.h:
(secureArchiver): Added.
(secureArchiverFromMutableData): Deleted.

Source/WebKit:

* Shared/Cocoa/DataDetectionResult.mm:
(WebKit::DataDetectionResult::encode const):
* Shared/Cocoa/WebCoreArgumentCodersCocoa.mm:
(IPC::ArgumentCoder<WebCore::Payment>::encode):
(IPC::ArgumentCoder<WebCore::PaymentContact>::encode):
(IPC::ArgumentCoder<WebCore::PaymentMerchantSession>::encode):
(IPC::ArgumentCoder<WebCore::PaymentMethod>::encode):
* Shared/ios/InteractionInformationAtPosition.mm:
(WebKit::InteractionInformationAtPosition::encode const):
* Shared/mac/WebCoreArgumentCodersMac.mm:
(IPC::ArgumentCoder<ProtectionSpace>::encodePlatformData):
(IPC::ArgumentCoder<Credential>::encodePlatformData):
(IPC::ArgumentCoder<ContentFilterUnblockHandler>::encode):
(IPC::ArgumentCoder<MediaPlaybackTargetContext>::encodePlatformData):
* Shared/mac/WebHitTestResultData.mm:
(WebKit::WebHitTestResultData::platformEncode const):
* UIProcess/API/Cocoa/WKProcessPool.mm:
(-[WKProcessPool _setObject:forBundleParameter:]):
(-[WKProcessPool _setObjectsForBundleParametersWithDictionary:]):
* UIProcess/Cocoa/WebProcessPoolCocoa.mm:
(WebKit::WebProcessPool::platformInitializeWebProcess):
* WebProcess/InjectedBundle/API/mac/WKWebProcessPlugInBrowserContextController.mm:
(-[WKWebProcessPlugInBrowserContextController _setFormDelegate:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225309 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoServiceWorkerClient objects should be reused if there is already one existing with...
commit-queue@webkit.org [Thu, 30 Nov 2017 01:19:25 +0000 (01:19 +0000)]
ServiceWorkerClient objects should be reused if there is already one existing with the same identifier
https://bugs.webkit.org/show_bug.cgi?id=180143

Patch by Youenn Fablet <youenn@apple.com> on 2017-11-29
Reviewed by Chris Dumez.

Source/WebCore:

Covered by updated tests.

ServiceWorkerGlobalScope keeps a map of all live ServiceWorkerClient objects.
Before creating a new client, it checks whether the map has one such object with the same identifier.
If so, it reuses this object. Otherwise it creates either a ServiceWorkerWindowClient or ServiceWorkerClient.

Add support for using a ServiceWorkerClientIdentifier as a HashMap key.

* workers/service/ServiceWorkerClient.cpp:
(WebCore::ServiceWorkerClient::ServiceWorkerClient):
(WebCore::ServiceWorkerClient::~ServiceWorkerClient):
* workers/service/ServiceWorkerClient.h:
(WebCore::ServiceWorkerClient::getOrCreate):
* workers/service/ServiceWorkerClientIdentifier.h:
(WebCore::ServiceWorkerClientIdentifier::hash const):
(WTF::ServiceWorkerClientIdentifierHash::hash):
(WTF::ServiceWorkerClientIdentifierHash::equal):
(WTF::HashTraits<WebCore::ServiceWorkerClientIdentifier>::emptyValue):
(WTF::HashTraits<WebCore::ServiceWorkerClientIdentifier>::constructDeletedValue):
(WTF::HashTraits<WebCore::ServiceWorkerClientIdentifier>::isDeletedValue):
* workers/service/ServiceWorkerGlobalScope.cpp:
(WebCore::ServiceWorkerGlobalScope::serviceWorkerClient):
(WebCore::ServiceWorkerGlobalScope::addServiceWorkerClient):
(WebCore::ServiceWorkerGlobalScope::removeServiceWorkerClient):
* workers/service/ServiceWorkerGlobalScope.h:
* workers/service/ServiceWorkerWindowClient.cpp:
(WebCore::ServiceWorkerWindowClient::ServiceWorkerWindowClient):
* workers/service/ServiceWorkerWindowClient.h:
* workers/service/context/ServiceWorkerThread.cpp:
(WebCore::ServiceWorkerThread::postMessageToServiceWorkerGlobalScope):

LayoutTests:

* http/tests/workers/service/resources/basic-ServiceWorker-postMessage-worker.js:
(event.else):
* http/tests/workers/service/resources/basic-ServiceWorker-postMessage.js:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225308 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoRemove pointer caging for double arrays
sbarati@apple.com [Thu, 30 Nov 2017 01:05:01 +0000 (01:05 +0000)]
Remove pointer caging for double arrays
https://bugs.webkit.org/show_bug.cgi?id=180163

Reviewed by Mark Lam.

This patch removes pointer caging from double arrays. Like
my previous removals of pointer caging, this is a security vs
performance tradeoff. We believe that butterflies being allocated
in the cage and with a 32GB runway gives us enough security that
pointer caging the butterfly just for double arrays does not add
enough security benefit for the performance hit it incurs.

This patch also removes the GetButterflyWithoutCaging node and
the FixedButterflyAccessUncaging phase. The node is no longer needed
because now all GetButterfly nodes are not caged. The phase is removed
since we no longer have two nodes.

* dfg/DFGAbstractInterpreterInlines.h:
(JSC::DFG::AbstractInterpreter<AbstractStateType>::executeEffects):
* dfg/DFGArgumentsEliminationPhase.cpp:
* dfg/DFGClobberize.h:
(JSC::DFG::clobberize):
* dfg/DFGDoesGC.cpp:
(JSC::DFG::doesGC):
* dfg/DFGFixedButterflyAccessUncagingPhase.cpp: Removed.
* dfg/DFGFixedButterflyAccessUncagingPhase.h: Removed.
* dfg/DFGFixupPhase.cpp:
(JSC::DFG::FixupPhase::fixupNode):
* dfg/DFGHeapLocation.cpp:
(WTF::printInternal):
* dfg/DFGHeapLocation.h:
* dfg/DFGNodeType.h:
* dfg/DFGPlan.cpp:
(JSC::DFG::Plan::compileInThreadImpl):
* dfg/DFGPredictionPropagationPhase.cpp:
* dfg/DFGSafeToExecute.h:
(JSC::DFG::safeToExecute):
* dfg/DFGSpeculativeJIT.cpp:
(JSC::DFG::SpeculativeJIT::compileSpread):
(JSC::DFG::SpeculativeJIT::compileArraySlice):
(JSC::DFG::SpeculativeJIT::compileGetButterfly):
* dfg/DFGSpeculativeJIT32_64.cpp:
(JSC::DFG::SpeculativeJIT::compile):
* dfg/DFGSpeculativeJIT64.cpp:
(JSC::DFG::SpeculativeJIT::compile):
* dfg/DFGTypeCheckHoistingPhase.cpp:
(JSC::DFG::TypeCheckHoistingPhase::identifyRedundantStructureChecks):
(JSC::DFG::TypeCheckHoistingPhase::identifyRedundantArrayChecks):
* ftl/FTLCapabilities.cpp:
(JSC::FTL::canCompile):
* ftl/FTLLowerDFGToB3.cpp:
(JSC::FTL::DFG::LowerDFGToB3::compileNode):
(JSC::FTL::DFG::LowerDFGToB3::compileGetButterfly):
* jit/JITPropertyAccess.cpp:
(JSC::JIT::emitDoubleLoad):
(JSC::JIT::emitGenericContiguousPutByVal):
* runtime/Butterfly.h:
(JSC::Butterfly::pointer):
(JSC::Butterfly::contiguousDouble):
(JSC::Butterfly::caged): Deleted.
* runtime/ButterflyInlines.h:
(JSC::Butterfly::createOrGrowPropertyStorage):
* runtime/JSObject.cpp:
(JSC::JSObject::ensureLengthSlow):
(JSC::JSObject::reallocateAndShrinkButterfly):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225307 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoMark imported/w3c/web-platform-tests/XMLHttpRequest/firing-events-http-no-content...
ryanhaddad@apple.com [Thu, 30 Nov 2017 00:39:14 +0000 (00:39 +0000)]
Mark imported/w3c/web-platform-tests/XMLHttpRequest/firing-events-http-no-content-length.html as flaky.
https://bugs.webkit.org/show_bug.cgi?id=179775

Unreviewed test gardening.

* platform/mac-wk1/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225306 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoMark media/modern-media-controls/pip-support/pip-support-click.html as flaky.
ryanhaddad@apple.com [Thu, 30 Nov 2017 00:39:12 +0000 (00:39 +0000)]
Mark media/modern-media-controls/pip-support/pip-support-click.html as flaky.
https://bugs.webkit.org/show_bug.cgi?id=165311

Unreviewed test gardening.

* platform/mac/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225305 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoMark http/tests/workers/service/registration-clear-redundant-worker.html as flaky.
ryanhaddad@apple.com [Thu, 30 Nov 2017 00:39:10 +0000 (00:39 +0000)]
Mark http/tests/workers/service/registration-clear-redundant-worker.html as flaky.
https://bugs.webkit.org/show_bug.cgi?id=179778

Unreviewed test gardening.

* TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225304 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoAdd missing WTF_MAKE_ISO_ALLOCATED macros
zalan@apple.com [Thu, 30 Nov 2017 00:31:50 +0000 (00:31 +0000)]
Add missing WTF_MAKE_ISO_ALLOCATED macros
https://bugs.webkit.org/show_bug.cgi?id=180165
<rdar://problem/35755617>

Reviewed by Simon Fraser.

* rendering/RenderFrame.h:
* rendering/RenderFullScreen.cpp:
* rendering/RenderIFrame.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225303 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoAPI test fix after r225288.
simon.fraser@apple.com [Thu, 30 Nov 2017 00:18:05 +0000 (00:18 +0000)]
API test fix after r225288.

Make the test work on iOS.

* TestWebKitAPI/Tests/WebKit/NoHistoryItemScrollToFragment.mm:
(-[DidScrollToFragmentScrollViewDelegate scrollViewDidScroll:]):
(TestWebKitAPI::TEST):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225302 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoSource/JavaScriptCore:
clopez@igalia.com [Thu, 30 Nov 2017 00:16:19 +0000 (00:16 +0000)]
Source/JavaScriptCore:
[MIPS][JSC] Implement MacroAssembler::probe support on MIPS
https://bugs.webkit.org/show_bug.cgi?id=175447

Patch by Stanislav Ocovaj <stanislav.ocovaj@rt-rk.com> on 2017-11-29
Reviewed by Carlos Alberto Lopez Perez.

This patch allows DFG JIT to be enabled on MIPS platforms.

* Sources.txt:
* assembler/MIPSAssembler.h:
(JSC::MIPSAssembler::lastSPRegister):
(JSC::MIPSAssembler::numberOfSPRegisters):
(JSC::MIPSAssembler::sprName):
* assembler/MacroAssemblerMIPS.cpp: Added.
(JSC::MacroAssembler::probe):
* assembler/ProbeContext.cpp:
(JSC::Probe::executeProbe):
* assembler/ProbeContext.h:
(JSC::Probe::CPUState::pc):
* assembler/testmasm.cpp:
(JSC::isSpecialGPR):
(JSC::testProbePreservesGPRS):
(JSC::testProbeModifiesStackPointer):
(JSC::testProbeModifiesStackValues):

Source/WTF:
[DFG][MIPS] Enable DFG JIT on MIPS.
https://bugs.webkit.org/show_bug.cgi?id=175447

Patch by Stanislav Ocovaj <stanislav.ocovaj@rt-rk.com> on 2017-11-29
Reviewed by Carlos Alberto Lopez Perez.

* wtf/Platform.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225301 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoRemove the ImageSource from the class hierarchy that connects BitmapImage to ImageFrame
commit-queue@webkit.org [Wed, 29 Nov 2017 23:23:54 +0000 (23:23 +0000)]
Remove the ImageSource from the class hierarchy that connects BitmapImage to ImageFrame
https://bugs.webkit.org/show_bug.cgi?id=175595

Patch by Said Abou-Hallawa <sabouhallawa@apple.com> on 2017-11-29
Reviewed by Darin Adler.

Source/WebCore:

The class hierarchy that connects BitmapImage to ImageFrame has been
troublesome. ImageSource does not have a clear responsibility other than
a bridge that connects BitmapIamge to ImageFrameCache. Sharing the
ImageDecoder between ImageSource and ImageFrameCache is ugly and caused
few crashes in the past.

This patch will do the first step for fixing this issue. First get rid of
ImageSource by moving its APIs to ImageFrameCache and BitmapImage. Replace
all the instances of ImageSource by ImageFrameCache. The next step will
be to rename ImageFrameCache to ImageSource. But this will be done in a
follow-up patch

* Sources.txt:
* WebCore.xcodeproj/project.pbxproj:
* platform/graphics/BitmapImage.cpp:
(WebCore::BitmapImage::BitmapImage):
(WebCore::BitmapImage::~BitmapImage):
(WebCore::BitmapImage::destroyDecodedData):
(WebCore::BitmapImage::destroyDecodedDataIfNecessary):
(WebCore::BitmapImage::dataChanged):
(WebCore::BitmapImage::frameImageAtIndexCacheIfNeeded):
(WebCore::BitmapImage::draw):
(WebCore::BitmapImage::canUseAsyncDecodingForLargeImages const):
(WebCore::BitmapImage::shouldUseAsyncDecodingForAnimatedImages const):
(WebCore::BitmapImage::subsamplingLevelForScaleFactor):
(WebCore::BitmapImage::canDestroyDecodedData):
(WebCore::BitmapImage::internalStartAnimation):
(WebCore::BitmapImage::stopAnimation):
(WebCore::BitmapImage::decode):
(WebCore::BitmapImage::imageFrameAvailableAtIndex):
(WebCore::BitmapImage::dump const):
* platform/graphics/BitmapImage.h:
* platform/graphics/GraphicsContext3D.h:
* platform/graphics/ImageFrameCache.cpp:
(WebCore::ImageFrameCache::ImageFrameCache):
(WebCore::ImageFrameCache::ensureDecoderAvailable):
(WebCore::ImageFrameCache::setData):
(WebCore::ImageFrameCache::resetData):
(WebCore::ImageFrameCache::dataChanged):
(WebCore::ImageFrameCache::isAllDataReceived):
(WebCore::ImageFrameCache::clearFrameBufferCache):
(WebCore::ImageFrameCache::canUseAsyncDecoding):
(WebCore::ImageFrameCache::maximumSubsamplingLevel):
(WebCore::ImageFrameCache::setTargetContext):
(WebCore::ImageFrameCache::createFrameImageAtIndex):
(WebCore::ImageFrameCache::dump):
(WebCore::ImageFrameCache::setDecoder): Deleted.
(WebCore::ImageFrameCache::decoder const): Deleted.
* platform/graphics/ImageFrameCache.h:
(WebCore::ImageFrameCache::create):
(WebCore::ImageFrameCache::requestFrameAsyncDecodingAtIndex):
* platform/graphics/ImageSource.cpp: Removed.
* platform/graphics/ImageSource.h: Removed.
* platform/graphics/cairo/GraphicsContext3DCairo.cpp:
(WebCore::GraphicsContext3D::ImageExtractor::~ImageExtractor):
(WebCore::GraphicsContext3D::ImageExtractor::extractImage):
* platform/graphics/cg/GraphicsContext3DCG.cpp:
(WebCore::GraphicsContext3D::ImageExtractor::extractImage):
* platform/graphics/cg/ImageSourceCG.h:

Source/WebKitLegacy/mac:

* WebView/WebPreferences.mm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225300 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWeb Inspector: Styles Redesign: can't add new property after property without trailin...
nvasilyev@apple.com [Wed, 29 Nov 2017 23:16:20 +0000 (23:16 +0000)]
Web Inspector: Styles Redesign: can't add new property after property without trailing semicolon
https://bugs.webkit.org/show_bug.cgi?id=179587
<rdar://problem/35490858>

Reviewed by Timothy Hatcher.

* UserInterface/Models/CSSProperty.js:
(WI.CSSProperty.prototype._updateOwnerStyleText):
(WI.CSSProperty.prototype._appendSemicolonIfNeeded):
Add a semicolon before the new property if the last property doesn't have it already.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225299 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWhen managing context startups, make ServiceWorkerJobDataIdentifier's optional.
beidson@apple.com [Wed, 29 Nov 2017 23:16:03 +0000 (23:16 +0000)]
When managing context startups, make ServiceWorkerJobDataIdentifier's optional.
https://bugs.webkit.org/show_bug.cgi?id=180166

Reviewed by Chris Dumez.

Source/WebCore:

No new tests (No behavior change).

We'll sometimes be starting SW contexts without an associated job, so ServiceWorkerContextData
should not required actually having a job identifier.

* workers/service/ServiceWorkerContextData.h:
(WebCore::ServiceWorkerContextData::decode):

* workers/service/context/SWContextManager.h:

* workers/service/server/SWServer.cpp:
(WebCore::SWServer::scriptContextFailedToStart):
(WebCore::SWServer::scriptContextStarted):
(WebCore::SWServer::didFinishInstall):
* workers/service/server/SWServer.h:

* workers/service/server/SWServerToContextConnection.cpp:
(WebCore::SWServerToContextConnection::scriptContextFailedToStart):
(WebCore::SWServerToContextConnection::scriptContextStarted):
(WebCore::SWServerToContextConnection::didFinishInstall):
* workers/service/server/SWServerToContextConnection.h:

* workers/service/server/SWServerWorker.cpp:
(WebCore::SWServerWorker::scriptContextFailedToStart):
(WebCore::SWServerWorker::scriptContextStarted):
(WebCore::SWServerWorker::didFinishInstall):
* workers/service/server/SWServerWorker.h:

Source/WebKit:

* StorageProcess/ServiceWorker/WebSWServerToContextConnection.messages.in:

* WebProcess/Storage/WebSWContextManagerConnection.cpp:
(WebKit::WebSWContextManagerConnection::serviceWorkerStartedWithMessage):
(WebKit::WebSWContextManagerConnection::didFinishInstall):
* WebProcess/Storage/WebSWContextManagerConnection.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225298 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoAdd support for service worker generated redirections
commit-queue@webkit.org [Wed, 29 Nov 2017 23:03:49 +0000 (23:03 +0000)]
Add support for service worker generated redirections
https://bugs.webkit.org/show_bug.cgi?id=179498

Patch by Youenn Fablet <youenn@apple.com> on 2017-11-29
Reviewed by Darin Adler.

Source/WebCore:

Test: http/tests/workers/service/service-worker-redirection-fetch.https.html

Added redirection routines following fetch specification to:
- check whether a response is a redirection
- generate a redirected request from a request and its redirection response.
Added some specific redirect request generation following existing WebKit networking code.

* Modules/fetch/FetchResponse.cpp:
(WebCore::FetchResponse::redirect):
(WebCore::isRedirectStatus): Deleted.
* platform/network/ResourceRequestBase.cpp:
(WebCore::shouldUseGet):
(WebCore::ResourceRequestBase::redirectedRequest const):
* platform/network/ResourceRequestBase.h:
* platform/network/ResourceResponseBase.h:
(WebCore::ResourceResponseBase::isRedirectionStatusCode):
(WebCore::ResourceResponseBase::isRedirection const):

Source/WebKit:

Small refactoring to allow a service worker redirected fetch to follow the redirection through the service worker.

* WebProcess/Network/WebLoaderStrategy.cpp:
(WebKit::WebLoaderStrategy::scheduleLoad): Make use of the loader request in case a network load is needed after a service worker redirection.
* WebProcess/Storage/ServiceWorkerClientFetch.cpp:
(WebKit::ServiceWorkerClientFetch::create):
(WebKit::ServiceWorkerClientFetch::~ServiceWorkerClientFetch):
(WebKit::ServiceWorkerClientFetch::ServiceWorkerClientFetch):
(WebKit::ServiceWorkerClientFetch::start):
(WebKit::ServiceWorkerClientFetch::didReceiveResponse): Check for response.
Generate redirected request if needed and call loader callback to process the redirection.
Adding some states so that if didFinish is called before the willSendRequest callback, redirection is followed.
(WebKit::ServiceWorkerClientFetch::didFinish):
In case redirection should be followed, wait for didFinish to follow it.
This simplifies the model although introducing some limited latency.
* WebProcess/Storage/ServiceWorkerClientFetch.h:
* WebProcess/Storage/WebSWClientConnection.cpp:
(WebKit::WebSWClientConnection::startFetch):
* WebProcess/Storage/WebSWClientConnection.h:
* WebProcess/Storage/WebServiceWorkerProvider.cpp:
(WebKit::WebServiceWorkerProvider::handleFetch):

LayoutTests:

* http/tests/workers/service/resources/service-worker-redirection-fetch-worker.js: Added.
(event.event.request.url.indexOf):
(event.event.request.url.endsWith):
* http/tests/workers/service/service-worker-redirection-fetch.https-expected.txt: Added.
* http/tests/workers/service/service-worker-redirection-fetch.https.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225297 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoStart exposing self.registration inside service workers
cdumez@apple.com [Wed, 29 Nov 2017 22:19:23 +0000 (22:19 +0000)]
Start exposing self.registration inside service workers
https://bugs.webkit.org/show_bug.cgi?id=180162

Reviewed by Brady Eidson.

LayoutTests/imported/w3c:

* web-platform-tests/service-workers/service-worker/redirected-response.https-expected.txt:

Source/WebCore:

Start exposing self.registration inside service workers as per:
- https://w3c.github.io/ServiceWorker/#serviceworkerglobalscope-interface

This is very initial support:
- The operations on the registration (such as update) will reject the promise for now.
- The registration's service workers are not yet populated.

This will be implemented in a follow-up.

Tests: http/tests/workers/service/ServiceWorkerGlobalScope_registration_SameObject.html
       http/tests/workers/service/self_registration.html

* bindings/js/JSServiceWorkerGlobalScopeCustom.cpp:
(WebCore::JSServiceWorkerGlobalScope::visitAdditionalChildren):
* workers/service/ServiceWorkerContainer.cpp:
(WebCore::ServiceWorkerContainer::addRegistration):
* workers/service/ServiceWorkerContextData.cpp:
(WebCore::ServiceWorkerContextData::isolatedCopy const):
* workers/service/ServiceWorkerContextData.h:
(WebCore::ServiceWorkerContextData::encode const):
(WebCore::ServiceWorkerContextData::decode):
* workers/service/ServiceWorkerGlobalScope.cpp:
(WebCore::ServiceWorkerGlobalScope::ServiceWorkerGlobalScope):
* workers/service/ServiceWorkerGlobalScope.h:
(WebCore::ServiceWorkerGlobalScope::registration):
* workers/service/server/SWServer.cpp:
(WebCore::SWServer::updateWorker):
(WebCore::SWServer::installContextData):
* workers/service/server/SWServer.h:
* workers/service/server/SWServerJobQueue.cpp:
(WebCore::SWServerJobQueue::scriptFetchFinished):

LayoutTests:

Add layout test coverage.

* TestExpectations:
Skip test that covers self.registration.update() because it now times out. We do not support update()
on registrations inside service workers yet so the test times out waiting for the updatefound event
on the registration.

* http/tests/workers/service/ServiceWorkerGlobalScope_registration_SameObject-expected.txt: Added.
* http/tests/workers/service/ServiceWorkerGlobalScope_registration_SameObject.html: Added.
* http/tests/workers/service/resources/ServiceWorkerGlobalScope_registration_SameObject-worker.js: Added.
* http/tests/workers/service/resources/self_registration-worker.js: Added.
* http/tests/workers/service/self_registration-expected.txt: Added.
* http/tests/workers/service/self_registration.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225296 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoAdd support for FetchEvent.clientId
commit-queue@webkit.org [Wed, 29 Nov 2017 21:47:37 +0000 (21:47 +0000)]
Add support for FetchEvent.clientId
https://bugs.webkit.org/show_bug.cgi?id=180052

Patch by Youenn Fablet <youenn@apple.com> on 2017-11-29
Reviewed by Chris Dumez.

LayoutTests/imported/w3c:

* web-platform-tests/service-workers/service-worker/fetch-event.https-expected.txt:
* web-platform-tests/service-workers/service-worker/resources/clients-get-worker.js:
* web-platform-tests/service-workers/service-worker/resources/fetch-event-test-worker.js:

Source/WebCore:

Covered by updated test.

Add script execution context identifier as a FetchOptions parameter.
This is then sent to the service worker process which can then set FetchEvent.clientId appropriately.
If the fetch is for a subresource, clientId is used directly.
If it is a navigation, clientId is set to targetClientId.

* loader/FetchOptions.h:
(WebCore::FetchOptions::encode const):
(WebCore::FetchOptions::decode):
* loader/cache/CachedResourceLoader.cpp:
(WebCore::CachedResourceLoader::prepareFetch):
* loader/cache/CachedResourceRequest.cpp:
(WebCore::CachedResourceRequest::setClientIdentifierIfNeeded):
* loader/cache/CachedResourceRequest.h:
* workers/service/context/ServiceWorkerFetch.cpp:
(WebCore::ServiceWorkerFetch::dispatchFetchEvent):
* workers/service/context/ServiceWorkerFetch.h:
* workers/service/context/ServiceWorkerThread.cpp:
(WebCore::ServiceWorkerThread::postFetchTask):
* workers/service/context/ServiceWorkerThread.h:

Source/WebKit:

Using FetchOption persistency coders for cache API and modernizing IPC FetchOptions decoding.

* WebKit/Shared/WebCoreArgumentCoders.cpp:
* NetworkProcess/cache/CacheStorageEngineCache.cpp:
(WebKit::CacheStorage::Cache::encode):
(WebKit::CacheStorage::Cache::decodeRecordHeader):
* WebProcess/Storage/WebSWContextManagerConnection.cpp:
(WebKit::WebSWContextManagerConnection::startFetch):
* WebProcess/Storage/WebSWContextManagerConnection.cpp:
(WebKit::WebSWContextManagerConnection::startFetch):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225294 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoUnreviewed, rolling out r225286.
jlewis3@apple.com [Wed, 29 Nov 2017 21:42:00 +0000 (21:42 +0000)]
Unreviewed, rolling out r225286.

The source files within this patch have been marked as
executable.

Reverted changeset:

"[MIPS][JSC] Implement MacroAssembler::probe support on MIPS"
https://bugs.webkit.org/show_bug.cgi?id=175447
https://trac.webkit.org/changeset/225286

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225293 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoMake WebFrameLoaderClient more robust against null pointer dereferencing
achristensen@apple.com [Wed, 29 Nov 2017 21:41:59 +0000 (21:41 +0000)]
Make WebFrameLoaderClient more robust against null pointer dereferencing
https://bugs.webkit.org/show_bug.cgi?id=180157
<rdar://problem/34895616>

Reviewed by Tim Horton.

There has always been rare null pointer crashes in this code, but they have become more common
now that we are waiting for completion handlers for redirects, which makes it more likely that
we are hitting this code after we have detached from the core frame.

* WebProcess/WebCoreSupport/WebFrameLoaderClient.cpp:
(WebKit::WebFrameLoaderClient::dispatchDecidePolicyForResponse):
(WebKit::WebFrameLoaderClient::dispatchDecidePolicyForNewWindowAction):
(WebKit::WebFrameLoaderClient::dispatchDecidePolicyForNavigationAction):
* WebProcess/WebPage/WebFrame.cpp:
(WebKit::WebFrame::page const):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225292 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoFlaky crash in WebCore::DOMGuardedObject::clear() during service worker tests
commit-queue@webkit.org [Wed, 29 Nov 2017 21:39:56 +0000 (21:39 +0000)]
Flaky crash in WebCore::DOMGuardedObject::clear() during service worker tests
https://bugs.webkit.org/show_bug.cgi?id=180045
<rdar://problem/35737288>

Patch by Youenn Fablet <youenn@apple.com> on 2017-11-29
Reviewed by Chris Dumez.

Manually tested by running concurrently service worker tests using FetchEvents which store promise references.

Before the patch, on workers, clearing of DOMGuardedObjects happens at the time WorkerGlobalScope is destroyed.
This is too late as it is expected that the JSDOMGlobalObject is still alive.

This patch adds a clearDOMGuardedObjects method on JSWorkerGlobalScopeBase.
It is called when stopping a WorkerThread, just before releasing the strong reference to JSWorkerGlobalScopeBase.

* bindings/js/JSDOMGuardedObject.h:
* bindings/js/JSWorkerGlobalScopeBase.cpp:
(WebCore::JSWorkerGlobalScopeBase::clearDOMGuardedObjects):
* bindings/js/JSWorkerGlobalScopeBase.h:
* bindings/js/WorkerScriptController.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225291 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoFix Mac CMake build.
achristensen@apple.com [Wed, 29 Nov 2017 21:31:07 +0000 (21:31 +0000)]
Fix Mac CMake build.

.:

* Source/cmake/OptionsMac.cmake:

Source/JavaScriptCore:

* PlatformMac.cmake:

Source/WebCore:

* PlatformMac.cmake:

Source/WebCore/PAL:

* pal/PlatformMac.cmake:

Source/WebKit:

* PlatformMac.cmake:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225290 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoStorageToWebProcessConnection & WebSWServerToContextConnection should use the same...
cdumez@apple.com [Wed, 29 Nov 2017 21:11:04 +0000 (21:11 +0000)]
StorageToWebProcessConnection & WebSWServerToContextConnection should use the same underlying IPC::Connection
https://bugs.webkit.org/show_bug.cgi?id=180147

Reviewed by Brady Eidson.

StorageToWebProcessConnection & WebSWServerToContextConnection should use the same underlying IPC::Connection.
Otherwise, we have with 2 IPC::Connections between the StorageProcess and the ServiceWorker (aka Context) process,
which makes synchronization of IPC messages difficult.

* StorageProcess/StorageProcess.cpp:
(WebKit::StorageProcess::createStorageToWebProcessConnection):
(WebKit::StorageProcess::createServerToContextConnection):
* StorageProcess/StorageProcess.h:
* StorageProcess/StorageProcess.messages.in:
* StorageProcess/StorageToWebProcessConnection.cpp:
(WebKit::StorageToWebProcessConnection::didReceiveMessage):
* UIProcess/ServiceWorkerProcessProxy.cpp:
(WebKit::ServiceWorkerProcessProxy::start):
* UIProcess/ServiceWorkerProcessProxy.h:
* UIProcess/Storage/StorageProcessProxy.cpp:
(WebKit::StorageProcessProxy::getStorageProcessConnection):
(WebKit::StorageProcessProxy::didFinishLaunching):
(WebKit::StorageProcessProxy::establishWorkerContextConnectionToStorageProcess):
* UIProcess/Storage/StorageProcessProxy.h:
* UIProcess/Storage/StorageProcessProxy.messages.in:
* UIProcess/WebProcessPool.cpp:
(WebKit::WebProcessPool::getStorageProcessConnection):
(WebKit::WebProcessPool::establishWorkerContextConnectionToStorageProcess):
* UIProcess/WebProcessPool.h:
* UIProcess/WebProcessProxy.cpp:
(WebKit::WebProcessProxy::getStorageProcessConnection):
* UIProcess/WebProcessProxy.h:
(WebKit::WebProcessProxy::isServiceWorkerProcess const):
* UIProcess/WebProcessProxy.messages.in:
* WebProcess/Storage/WebToStorageProcessConnection.cpp:
(WebKit::WebToStorageProcessConnection::didReceiveMessage):
* WebProcess/WebProcess.cpp:
(WebKit::WebProcess::didReceiveMessage):
(WebKit::WebProcess::establishWorkerContextConnectionToStorageProcess):
* WebProcess/WebProcess.h:
* WebProcess/WebProcess.messages.in:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225289 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoREGRESSION (r213590): Scrolling to anchors broken in web views when content loaded...
simon.fraser@apple.com [Wed, 29 Nov 2017 20:41:47 +0000 (20:41 +0000)]
REGRESSION (r213590): Scrolling to anchors broken in web views when content loaded via HTML string
https://bugs.webkit.org/show_bug.cgi?id=180155
rdar://problem/34220827

Reviewed by Zalan Bujtas.
Source/WebCore:

When content is loaded in a UIWebView or WKWebView via an HTML string, history().currentItem()
is null so itemAllowsScrollRestoration() would return false, preventing scrolling to anchors.

Fix by allowing scroll restoration if the the history item is null.

Tested by WebKit.NoHistoryItemScrollToFragment API test.

* loader/FrameLoader.cpp:
(WebCore::itemAllowsScrollRestoration):

Tools:

API test that loads a page with a relative anchor, and simulates a click to scroll to it.

* TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
* TestWebKitAPI/Tests/WebKit/MediaStreamTrackDetached.mm:
(TestWebKitAPI::TEST):
* TestWebKitAPI/Tests/WebKit/NoHistoryItemScrollToFragment.mm: Added.
(-[DidScrollToFragmentDelegate _webViewDidScroll:]):
(TestWebKitAPI::TEST):
* TestWebKitAPI/Tests/WebKit/scroll-to-anchor.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225288 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agowebkitpy EWS should be present on the dashboard.
ryanhaddad@apple.com [Wed, 29 Nov 2017 20:30:37 +0000 (20:30 +0000)]
webkitpy EWS should be present on the dashboard.
https://bugs.webkit.org/show_bug.cgi?id=179801

Reviewed by Aakash Jain.

* BuildSlaveSupport/build.webkit.org-config/public_html/dashboard/Scripts/BubbleQueueServer.js:
(BubbleQueueServer): Add Webkitpy EWS.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225287 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoSource/JavaScriptCore:
commit-queue@webkit.org [Wed, 29 Nov 2017 20:12:32 +0000 (20:12 +0000)]
Source/JavaScriptCore:
[MIPS][JSC] Implement MacroAssembler::probe support on MIPS
https://bugs.webkit.org/show_bug.cgi?id=175447

Patch by Stanislav Ocovaj <stanislav.ocovaj@rt-rk.com> on 2017-11-29
Reviewed by Carlos Alberto Lopez Perez.

This patch allows DFG JIT to be enabled on MIPS platforms.

* Sources.txt:
* assembler/MIPSAssembler.h:
(JSC::MIPSAssembler::lastSPRegister):
(JSC::MIPSAssembler::numberOfSPRegisters):
(JSC::MIPSAssembler::sprName):
* assembler/MacroAssemblerMIPS.cpp: Added.
(JSC::MacroAssembler::probe):
* assembler/ProbeContext.cpp:
(JSC::Probe::executeProbe):
* assembler/ProbeContext.h:
(JSC::Probe::CPUState::pc):
* assembler/testmasm.cpp:
(JSC::isSpecialGPR):
(JSC::testProbePreservesGPRS):
(JSC::testProbeModifiesStackPointer):
(JSC::testProbeModifiesStackValues):

Source/WTF:
[DFG][MIPS] Enable DFG JIT on MIPS.
https://bugs.webkit.org/show_bug.cgi?id=175447

Patch by Stanislav Ocovaj <stanislav.ocovaj@rt-rk.com> on 2017-11-29
Reviewed by Carlos Alberto Lopez Perez.

* wtf/Platform.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225286 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoIntroduce ServiceWorkerContainer::ensureSWClientConnection()
cdumez@apple.com [Wed, 29 Nov 2017 19:41:22 +0000 (19:41 +0000)]
Introduce ServiceWorkerContainer::ensureSWClientConnection()
https://bugs.webkit.org/show_bug.cgi?id=180146

Reviewed by Youenn Fablet.

Introduce ServiceWorkerContainer::ensureSWClientConnection() to reduce
code duplication. Also use callOnMainThread() in preparation for this
getting called from a service worker thread (now that ServiceWorkerContainer
is exposed to service workers). This is needed because constructing the
SWClientConnection initializes the IPC connection from the WebProcess to
the StorageProcess, which involves a synchronous IPC with the UIProcess.
Doing a synchronous IPC from a background thread is unsupported.

* workers/service/ServiceWorkerContainer.cpp:
(WebCore::ServiceWorkerContainer::addRegistration):
(WebCore::ServiceWorkerContainer::getRegistration):
(WebCore::ServiceWorkerContainer::getRegistrations):
(WebCore::ServiceWorkerContainer::ensureSWClientConnection):
* workers/service/ServiceWorkerContainer.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225283 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoensure*Connection() methods on WebProcess should return a reference
cdumez@apple.com [Wed, 29 Nov 2017 19:40:12 +0000 (19:40 +0000)]
ensure*Connection() methods on WebProcess should return a reference
https://bugs.webkit.org/show_bug.cgi?id=180149

Reviewed by Alex Christensen.

ensure*Connection() methods on WebProcess should return a reference instead of not returning
anything. Also get rid of the non-ensure variants which called "ensure" internally and are
no longer needed.

* Shared/mac/CookieStorageShim.mm:
(WebKit::webKitCookieStorageCopyRequestHeaderFieldsForURL):
* WebProcess/Cache/WebCacheStorageConnection.cpp:
(WebKit::WebCacheStorageConnection::connection):
* WebProcess/Databases/IndexedDB/WebIDBConnectionToServer.cpp:
(WebKit::WebIDBConnectionToServer::messageSenderConnection):
(WebKit::preregisterSandboxExtensionsIfNecessary):
* WebProcess/Databases/WebDatabaseProvider.cpp:
(WebKit::WebDatabaseProvider::idbConnectionToServerForSession):
* WebProcess/FileAPI/BlobRegistryProxy.cpp:
(WebKit::BlobRegistryProxy::registerFileBlobURL):
(WebKit::BlobRegistryProxy::registerBlobURL):
(WebKit::BlobRegistryProxy::registerBlobURLOptionallyFileBacked):
(WebKit::BlobRegistryProxy::unregisterBlobURL):
(WebKit::BlobRegistryProxy::registerBlobURLForSlice):
(WebKit::BlobRegistryProxy::blobSize):
(WebKit::BlobRegistryProxy::writeBlobsToTemporaryFiles):
* WebProcess/Network/NetworkProcessConnection.cpp:
(WebKit::NetworkProcessConnection::writeBlobsToTemporaryFiles):
* WebProcess/Network/WebLoaderStrategy.cpp:
(WebKit::WebLoaderStrategy::scheduleLoadFromNetworkProcess):
(WebKit::WebLoaderStrategy::remove):
(WebKit::WebLoaderStrategy::setDefersLoading):
(WebKit::WebLoaderStrategy::loadResourceSynchronously):
(WebKit::WebLoaderStrategy::startPingLoad):
(WebKit::WebLoaderStrategy::preconnectTo):
(WebKit::WebLoaderStrategy::storeDerivedDataToCache):
(WebKit::WebLoaderStrategy::setCaptureExtraNetworkLoadMetricsEnabled):
* WebProcess/Network/WebResourceLoader.cpp:
(WebKit::WebResourceLoader::messageSenderConnection):
* WebProcess/Network/WebSocketStream.cpp:
(WebKit::WebSocketStream::WebSocketStream):
(WebKit::WebSocketStream::messageSenderConnection):
* WebProcess/Network/webrtc/LibWebRTCResolver.cpp:
(WebKit::sendOnMainThread):
* WebProcess/Network/webrtc/LibWebRTCSocket.cpp:
(WebKit::sendOnMainThread):
* WebProcess/Network/webrtc/LibWebRTCSocketFactory.cpp:
(WebKit::LibWebRTCSocketFactory::CreateServerTcpSocket):
(WebKit::LibWebRTCSocketFactory::CreateUdpSocket):
(WebKit::LibWebRTCSocketFactory::CreateClientTcpSocket):
(WebKit::LibWebRTCSocketFactory::createNewConnectionSocket):
* WebProcess/Network/webrtc/WebRTCMonitor.cpp:
(WebKit::sendOnMainThread):
* WebProcess/Storage/WebServiceWorkerProvider.cpp:
(WebKit::WebServiceWorkerProvider::serviceWorkerConnectionForSession):
(WebKit::WebServiceWorkerProvider::handleFetch):
* WebProcess/WebCoreSupport/WebPlatformStrategies.cpp:
(WebKit::WebPlatformStrategies::cookiesForDOM):
(WebKit::WebPlatformStrategies::setCookiesFromDOM):
(WebKit::WebPlatformStrategies::cookiesEnabled):
(WebKit::WebPlatformStrategies::cookieRequestHeaderFieldValue):
(WebKit::WebPlatformStrategies::getRawCookies):
(WebKit::WebPlatformStrategies::deleteCookie):
* WebProcess/WebPage/WebFrame.cpp:
(WebKit::WebFrame::startDownload):
(WebKit::WebFrame::convertMainResourceLoadToDownload):
* WebProcess/WebProcess.cpp:
(WebKit::WebProcess::ensureLegacyPrivateBrowsingSessionInNetworkProcess):
(WebKit::WebProcess::ensureNetworkProcessConnection):
(WebKit::WebProcess::ensureWebToStorageProcessConnection):
(WebKit::WebProcess::prefetchDNS):
* WebProcess/WebProcess.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225282 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[GTK] Test gardening
commit-queue@webkit.org [Wed, 29 Nov 2017 19:22:24 +0000 (19:22 +0000)]
[GTK] Test gardening
https://bugs.webkit.org/show_bug.cgi?id=180150

Unreviewed test gardening

Patch by Javier M. Mellid <jmunhoz@igalia.com> on 2017-11-29

* platform/gtk/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225281 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWeb Inspector: Console Tab navigation bar sometimes does not include filter bar,...
joepeck@webkit.org [Wed, 29 Nov 2017 19:07:11 +0000 (19:07 +0000)]
Web Inspector: Console Tab navigation bar sometimes does not include filter bar, clear console sometimes does not work
https://bugs.webkit.org/show_bug.cgi?id=180124
<rdar://problem/35740353>

Reviewed by Brian Burg.

* UserInterface/Views/LogContentView.js:
(WI.LogContentView.prototype.closed):
Avoid removing event listeners if this LogContentView singleton is ever closed.
The singleton will always be alive so we don't want to remove the event listeners
without a way to add them back.

* UserInterface/Base/Main.js:
(WI.showSplitConsole):
* UserInterface/Views/ConsoleTabContentView.js:
(WI.ConsoleTabContentView.prototype.shown):
When showing the ConsoleTab immediately collapse the split console so that any
following code that checks WI.isShowingSplitConsole will get the expected value.
It is also now possible to share a ContentView across ContentBrowsers via
tombstones, so remove the old code that would frequently close the LogContentView.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225280 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[iOS] Media controls should stop updating while media is playing in fullscreen
commit-queue@webkit.org [Wed, 29 Nov 2017 19:05:10 +0000 (19:05 +0000)]
[iOS] Media controls should stop updating while media is playing in fullscreen
https://bugs.webkit.org/show_bug.cgi?id=180144
<rdar://problem/35060379>

Patch by Antoine Quint <graouts@apple.com> on 2017-11-29
Reviewed by Eric Carlson.

Source/WebCore:

Updating inline media controls while playing media in fullscreen is useless since we're guaranteed not to
have those controls visible, and hurtful since this has impact on battery life. To avoid this, we remove
all media event listeners while in fullscreen on iOS, which will prevent the UI to be udpated since all
updates are driven by media events.

To implement this, we remove the MediaControllerSupport destroy() method and make it a disable() method,
and factor code out of the MediaControllerSupport constructor into an enable() method that registers the
media event listeners. Then, as we enter and exit fullscreen, we call the disable() and enable() method
on the various MediaControllerSupport objects that were created to support the iOS inline media controls.

Test: media/modern-media-controls/media-controller/ios/media-controller-stop-updates-in-fullscreen.html

* Modules/modern-media-controls/media/controls-visibility-support.js:
(ControlsVisibilitySupport):
(ControlsVisibilitySupport.prototype.enable):
(ControlsVisibilitySupport.prototype.disable):
(ControlsVisibilitySupport.prototype.destroy): Deleted.
* Modules/modern-media-controls/media/media-controller-support.js:
(MediaControllerSupport):
(MediaControllerSupport.prototype.enable):
(MediaControllerSupport.prototype.disable):
(MediaControllerSupport.prototype.destroy): Deleted.
* Modules/modern-media-controls/media/media-controller.js:
(MediaController.prototype.handleEvent):
(MediaController.prototype._updateControlsIfNeeded):
(MediaController.prototype._updateSupportingObjectsEnabledState):
(MediaController):

LayoutTests:

Add a new test that enters fullscreen, checks that the elapsed time shown in the inline media controls are the same
at this time and after a few "timeupdate" events, ensuring the DOM is no longer updated while in fullscreen, and then
exits fullscreen and checks that the elapsed time controls now update as expected. This test is skipped in OpenSource
since it uses touch events.

* media/modern-media-controls/media-controller/ios/media-controller-stop-updates-in-fullscreen-expected.txt: Added.
* media/modern-media-controls/media-controller/ios/media-controller-stop-updates-in-fullscreen.html: Added.
* platform/ios-simulator/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225279 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoModernize API::SerializedScriptValue
commit-queue@webkit.org [Wed, 29 Nov 2017 19:03:23 +0000 (19:03 +0000)]
Modernize API::SerializedScriptValue
https://bugs.webkit.org/show_bug.cgi?id=180115

Patch by Alex Christensen <achristensen@webkit.org> on 2017-11-29
Reviewed by Brady Eidson.

Also remove some SPI that hasn't been used anywhere since Mountain Lion.

* Shared/API/APISerializedScriptValue.h:
(API::SerializedScriptValue::create):
(API::SerializedScriptValue::internalRepresentation):
(API::SerializedScriptValue::SerializedScriptValue):
* Shared/API/c/WKSerializedScriptValue.cpp:
(WKSerializedScriptValueDeserialize):
(WKSerializedScriptValueCreateWithInternalRepresentation): Deleted.
(WKSerializedScriptValueGetInternalRepresentation): Deleted.
* Shared/API/c/WKSerializedScriptValuePrivate.h: Removed.
* UIProcess/API/Cocoa/WKWebView.mm:
(-[WKWebView _evaluateJavaScript:forceUserGesture:completionHandler:]):
* WebKit.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225278 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoViewport unit values affected by Comand-+ zoom
simon.fraser@apple.com [Wed, 29 Nov 2017 18:59:39 +0000 (18:59 +0000)]
Viewport unit values affected by Comand-+ zoom
https://bugs.webkit.org/show_bug.cgi?id=145614

Reviewed by Zalan Bujtas.
Source/WebCore:

Don't apply zooming when resolving viewport-relative lengths, since they should not
change based on the zoom level.

Test: fast/css/viewport-units-zoom.html

* css/CSSPrimitiveValue.cpp:
(WebCore::CSSPrimitiveValue::computeNonCalcLengthDouble):

Tools:

Fix Command+ and Command- in MiniBrowser to do zooming, rather than being tied
to editing commands.

* MiniBrowser/mac/MainMenu.xib:

LayoutTests:

* fast/css/viewport-units-zoom-expected.html: Added.
* fast/css/viewport-units-zoom.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225277 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoUnreviewed test fix after r225264.
bfulgham@apple.com [Wed, 29 Nov 2017 18:59:01 +0000 (18:59 +0000)]
Unreviewed test fix after r225264.
<rdar://problem/35750689>

The changes in r225264 were meant to have no changes in behavior. However, I mistakenly
switched to a secure coding API call in the PlatformPasteboard::write method. This should
have used the 'insecure' version of this function.

* platform/ios/PlatformPasteboardIOS.mm:
(WebCore::PlatformPasteboard::write):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225276 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoAdd test for _WKVisitedLinkStore.addVisitedLinkWithString
achristensen@apple.com [Wed, 29 Nov 2017 18:53:11 +0000 (18:53 +0000)]
Add test for _WKVisitedLinkStore.addVisitedLinkWithString
https://bugs.webkit.org/show_bug.cgi?id=180152

Reviewed by Chris Dumez.

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

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225275 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agowebkitpy: Trying to use iOS versions from machines without iOS SDKs doesn't make...
jbedard@apple.com [Wed, 29 Nov 2017 18:04:06 +0000 (18:04 +0000)]
webkitpy: Trying to use iOS versions from machines without iOS SDKs doesn't make sense
https://bugs.webkit.org/show_bug.cgi?id=179534
<rdar://problem/35469509>

Reviewed by Brent Fulgham.

Provide more specific information in builders.py so that machines do not try and calculate
an iOS SDK version while running tests.

* Scripts/webkitpy/port/builders.py: Explicitly specify a version for iOS Simulator.
* Scripts/webkitpy/port/ios_simulator.py:
(IOSSimulatorPort._version_from_name): Attempt to extract the iOS version from the name
of the port.
(IOSSimulatorPort.ios_version): Use specified iOS version if a version is detected in
in the provided name string (just like Mac).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225274 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoStrict and sloppy functions shouldn't share structure
jfbastien@apple.com [Wed, 29 Nov 2017 17:46:26 +0000 (17:46 +0000)]
Strict and sloppy functions shouldn't share structure

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225273 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoLibWebRTCPeerConnectionBackend should clean its stats promises when being cleaned
commit-queue@webkit.org [Wed, 29 Nov 2017 17:45:19 +0000 (17:45 +0000)]
LibWebRTCPeerConnectionBackend should clean its stats promises when being cleaned
https://bugs.webkit.org/show_bug.cgi?id=180101

Patch by Youenn Fablet <youenn@apple.com> on 2017-11-29
Reviewed by Eric Carlson.

No change of behavior.

* Modules/mediastream/libwebrtc/LibWebRTCPeerConnectionBackend.cpp:
(WebCore::LibWebRTCPeerConnectionBackend::doStop): Cleaning stat promises hash map.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225272 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[JSC] Add MacroAssembler::getEffectiveAddress in all platforms
utatane.tea@gmail.com [Wed, 29 Nov 2017 17:37:36 +0000 (17:37 +0000)]
[JSC] Add MacroAssembler::getEffectiveAddress in all platforms
https://bugs.webkit.org/show_bug.cgi?id=180070

Reviewed by Saam Barati.

This patch adds getEffectiveAddress in all JIT platforms.
This is abstracted version of x86 lea.

We also fix a bug in Yarr that uses branch32 instead of branchPtr for addresses.

* assembler/MacroAssemblerARM.h:
(JSC::MacroAssemblerARM::getEffectiveAddress):
* assembler/MacroAssemblerARM64.h:
(JSC::MacroAssemblerARM64::getEffectiveAddress):
(JSC::MacroAssemblerARM64::getEffectiveAddress64): Deleted.
* assembler/MacroAssemblerARMv7.h:
(JSC::MacroAssemblerARMv7::getEffectiveAddress):
* assembler/MacroAssemblerMIPS.h:
(JSC::MacroAssemblerMIPS::getEffectiveAddress):
* assembler/MacroAssemblerX86.h:
(JSC::MacroAssemblerX86::getEffectiveAddress):
* assembler/MacroAssemblerX86_64.h:
(JSC::MacroAssemblerX86_64::getEffectiveAddress):
(JSC::MacroAssemblerX86_64::getEffectiveAddress64): Deleted.
* assembler/testmasm.cpp:
(JSC::testGetEffectiveAddress):
(JSC::run):
* dfg/DFGSpeculativeJIT.cpp:
(JSC::DFG::SpeculativeJIT::compileArrayPush):
* yarr/YarrJIT.cpp:
(JSC::Yarr::YarrGenerator::tryReadUnicodeCharImpl):
(JSC::Yarr::YarrGenerator::tryReadUnicodeChar):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225271 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoThe recursive tail call optimisation is wrong on closures
rmorisset@apple.com [Wed, 29 Nov 2017 17:31:54 +0000 (17:31 +0000)]
The recursive tail call optimisation is wrong on closures
https://bugs.webkit.org/show_bug.cgi?id=179835

Reviewed by Saam Barati.

JSTests:

* stress/closure-recursive-tail-call.js: Added.
(makeClosure):

PerformanceTests:

This new benchmark is a very close variant of the merge-sort benchmark, that writes mergeSorted in a kinda CPS style,
to stress the use of closures, and of polymorphic calls.

* TailBench9000/merge-sort-cps.js: Added.
(createRNG):
(mergeSorted):
(checkSorted.check):
(add):
(build):
(compare):
(checkSpectrum):
(buildArray):
(test):

Source/JavaScriptCore:

The problem is that we only check the executable of the callee, not whatever variables might have been captured.
As a stopgap measure this patch just does not do the optimisation for closures.

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

Tools:

This just includes merge-sort-cps.js to the list of benchmarks ran by run-jsc-benchmarks --tail-bench

* Scripts/run-jsc-benchmarks:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225270 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoLayoutTests/imported/w3c:
Ms2ger@igalia.com [Wed, 29 Nov 2017 17:11:36 +0000 (17:11 +0000)]
LayoutTests/imported/w3c:
Rebaseline imported/w3c/web-platform-tests/resource-timing/single-entry-per-resource.html.
https://bugs.webkit.org/show_bug.cgi?id=180142

Unreviewed test gardening.

* web-platform-tests/resource-timing/single-entry-per-resource-expected.txt:

LayoutTests:
Enable imported/w3c/web-platform-tests/resource-timing/single-entry-per-resource.html.
https://bugs.webkit.org/show_bug.cgi?id=180142

Unreviewed test gardening.

The -expected file was updated to list a PASS result.

* TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225269 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[GTK][WPE] Test gardening
Ms2ger@igalia.com [Wed, 29 Nov 2017 17:09:57 +0000 (17:09 +0000)]
[GTK][WPE] Test gardening
https://bugs.webkit.org/show_bug.cgi?id=180136

Unreviewed test gardening.

* TestExpectations: Skip more alternative-presentation-button tests.
* platform/gtk/TestExpectations: Updated expectations:
  - fast/attachment/attachment-without-appearance.html: recently added test for a disabled feature.
  - fast/dom/MutationObserver/end-of-task-delivery.html: flaky.
  - http/tests/local/link-stylesheet-load-order-preload.html: flaky.
* platform/gtk/compositing/overflow/composited-scrolling-paint-phases-expected.txt: rebaseline for r225220.
* platform/mac/TestExpectations: Enable the skipped alternative-presentation-button tests.
* platform/wpe/TestExpectations: Updated expectations:
  - fast/canvas/canvas-createPattern-video-modify.html: passing since r225060.
  - webanimations/*: failing since they were added; bug filed.
* platform/wpe/imported/w3c/web-platform-tests/dom/events/EventTarget-dispatchEvent-expected.txt:
  DeviceMotionEvent and DeviceMotionEvent were disabled in r225098
  (matching mac ports); the remaining difference with the platform-neutral
  expectation is that TouchEvent is enabled here.
* platform/wpe/imported/w3c/web-platform-tests/dom/nodes/Document-createEvent-expected.txt:
  DeviceMotionEvent and DeviceMotionEvent were disabled in r225098
  (matching mac ports); the remaining difference with the platform-neutral
  expectation is that TouchEvent is enabled here.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225268 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoREGRESSION(r218064): [GTK] Broke entering fullscreen mode in debug builds
mcatanzaro@igalia.com [Wed, 29 Nov 2017 17:06:58 +0000 (17:06 +0000)]
REGRESSION(r218064): [GTK] Broke entering fullscreen mode in debug builds
https://bugs.webkit.org/show_bug.cgi?id=180120

Reviewed by Carlos Garcia Campos.

These assertions need to be swapped. Fixes /webkit2/WebKitWebView/fullscreen in debug mode.

* UIProcess/API/gtk/WebKitWebViewBase.cpp:
(webkitWebViewBaseEnterFullScreen):
(webkitWebViewBaseExitFullScreen):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225267 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[CoordGraphics] Rename CoordinatedBuffer to Nicosia::Buffer
zandobersek@gmail.com [Wed, 29 Nov 2017 14:30:27 +0000 (14:30 +0000)]
[CoordGraphics] Rename CoordinatedBuffer to Nicosia::Buffer
https://bugs.webkit.org/show_bug.cgi?id=180135

Reviewed by Carlos Garcia Campos.

Source/WebCore:

Rename CoordinatedBuffer to Nicosia::Buffer, starting an abstraction
layer that will in the future allow us to prototype and potentially
support different 2D rasterization libraries. The layer is envisioned
as separate from the CoordinatedGraphics code, but will in the mid-term
only be used there.

In order to keep CMake changes to a minimum for now, the source code is
included in the build along with the CoordinatedGraphics source files,
in TextureMapper.cmake.

No new tests -- no change in functionality.

* platform/TextureMapper.cmake:
* platform/graphics/nicosia/NicosiaBuffer.cpp: Renamed from Source/WebCore/platform/graphics/texmap/coordinated/CoordinatedBuffer.cpp.
(Nicosia::Buffer::create):
(Nicosia::Buffer::Buffer):
(Nicosia::Buffer::context):
(Nicosia::Buffer::uploadImage):
* platform/graphics/nicosia/NicosiaBuffer.h: Renamed from Source/WebCore/platform/graphics/texmap/coordinated/CoordinatedBuffer.h.
(Nicosia::Buffer::size const):
* platform/graphics/texmap/coordinated/CoordinatedGraphicsLayer.cpp:
(WebCore::CoordinatedGraphicsLayer::updateContentBuffers):
* platform/graphics/texmap/coordinated/CoordinatedGraphicsLayer.h:
* platform/graphics/texmap/coordinated/CoordinatedGraphicsState.h:
* platform/graphics/texmap/coordinated/CoordinatedImageBacking.cpp:
(WebCore::CoordinatedImageBacking::update):
* platform/graphics/texmap/coordinated/CoordinatedImageBacking.h:
* platform/graphics/texmap/coordinated/Tile.h:

Source/WebKit:

Adjust code to the CoordinatedBuffer -> Nicosia::Buffer transition.

* Shared/CoordinatedGraphics/CoordinatedBackingStore.cpp:
(WebKit::CoordinatedBackingStoreTile::setBackBuffer):
(WebKit::CoordinatedBackingStore::updateTile):
* Shared/CoordinatedGraphics/CoordinatedBackingStore.h:
* Shared/CoordinatedGraphics/CoordinatedGraphicsScene.cpp:
(WebKit::CoordinatedGraphicsScene::createUpdateAtlas):
(WebKit::CoordinatedGraphicsScene::updateImageBacking):
* Shared/CoordinatedGraphics/CoordinatedGraphicsScene.h:
* WebProcess/WebPage/CoordinatedGraphics/CompositingCoordinator.cpp:
(WebKit::CompositingCoordinator::updateImageBacking):
(WebKit::CompositingCoordinator::createUpdateAtlas):
(WebKit::CompositingCoordinator::getCoordinatedBuffer):
* WebProcess/WebPage/CoordinatedGraphics/CompositingCoordinator.h:
* WebProcess/WebPage/CoordinatedGraphics/UpdateAtlas.cpp:
(WebKit::UpdateAtlas::UpdateAtlas):
(WebKit::UpdateAtlas::getCoordinatedBuffer):
* WebProcess/WebPage/CoordinatedGraphics/UpdateAtlas.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225266 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoPressing the space bar while watching a fullscreen video doesn't play or pause
graouts@webkit.org [Wed, 29 Nov 2017 13:31:36 +0000 (13:31 +0000)]
Pressing the space bar while watching a fullscreen video doesn't play or pause
https://bugs.webkit.org/show_bug.cgi?id=180033
<rdar://problem/33610443>

Reviewed by Eric Carlson.

Source/WebCore:

We register a "keydown" event to track when the space bar is pressed, and if the media is playing
in fullscreen, we toggle playback. This does not interfere with full keyboard access since activating
one of the media controls using the keyboard will not let the events we register for be dispatched
this far along the event dispatch phase.

Test: media/modern-media-controls/media-controller/media-controller-space-bar-toggle-playback.html

* Modules/modern-media-controls/media/media-controller.js:
(MediaController):
(MediaController.prototype.togglePlayback): Add a catch() statement since calling play() could sometime
lead to some extraneous unhandled promise console logging that pollutes test output.
(MediaController.prototype.handleEvent):

LayoutTests:

Adding a new macOS-only test that checks that pressing the space bar while playing fullscreen
pauses the media and resumes it when pressing the space bar again.

* media/modern-media-controls/media-controller/media-controller-space-bar-toggle-playback-expected.txt: Added.
* media/modern-media-controls/media-controller/media-controller-space-bar-toggle-playback.html: Added.
* media/video-test.js:
(runWithKeyDown): Update the key to not be space since this would cause media to be paused when entering fullscreen.
* platform/ios-simulator/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225265 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoAdopt updated NSKeyed[Un]Archiver API when available
bfulgham@apple.com [Wed, 29 Nov 2017 07:52:47 +0000 (07:52 +0000)]
Adopt updated NSKeyed[Un]Archiver API when available
https://bugs.webkit.org/show_bug.cgi?id=180127
<rdar://problem/35710738>

Reviewed by Alex Christensen.

Source/WebCore:

Switch to new NSKeyed[Un]Archiver methods when available. We do not attempt
to adopt secure coding in places we were not under the original API.

Most of the new API is wrapped in a set of convenience methods so we can
build without the new API on older systems.

No change in behavior.

* editing/cocoa/EditorCocoa.mm:
(WebCore::archivedDataForAttributedString): Use new convenience method
to archive the string object.
* loader/archive/cf/LegacyWebArchiveMac.mm:
(WebCore::LegacyWebArchive::createResourceResponseFromMacArchivedData): Update to
use secure coding where possible.
(WebCore::LegacyWebArchive::createPropertyListRepresentation): Ditto.
* platform/ios/PlatformPasteboardIOS.mm:
(WebCore::PlatformPasteboard::write): Use new secure API.
(WebCore::PlatformPasteboard::typesSafeForDOMToReadAndWrite const): Ditto.
* testing/cocoa/WebArchiveDumpSupport.mm:
(WebCoreTestSupport::createCFURLResponseFromResponseData): Update to
secure coding API where possible.

Source/WebCore/PAL:

Add new convenience methods (and SPI headers) so we can adopt new NSKeyedArchiver
API that uses secure coding by default.

* PAL.xcodeproj/project.pbxproj:
* pal/spi/cocoa/NSKeyedArchiverSPI.h: Added.
(securelyArchivedDataWithRootObject): New convenience method.
(securelyUnarchiveRootObjectOfClassFromData): Ditto.
(secureArchiverFromMutableData): Ditto.
(secureUnarchiverFromData): Ditto.

Source/WebKit:

Switch to new NSKeyed[Un]Archiver methods when available. We do not attempt
to adopt secure coding in places we were not under the original API.

* Platform/ios/AccessibilityIOS.mm:
(WebKit::newAccessibilityRemoteToken): Use secure-by-default API.
* Shared/Cocoa/DataDetectionResult.mm:
(WebKit::DataDetectionResult::encode const): Ditto.
(WebKit::DataDetectionResult::decode): Ditto.
* Shared/Cocoa/WebCoreArgumentCodersCocoa.mm:
(IPC::ArgumentCoder<WebCore::Payment>::encode): Ditto.
(IPC::ArgumentCoder<WebCore::Payment>::decode): Ditto.
(IPC::ArgumentCoder<WebCore::PaymentContact>::encode): Ditto.
(IPC::ArgumentCoder<WebCore::PaymentContact>::decode): Ditto.
(IPC::ArgumentCoder<WebCore::PaymentMerchantSession>::encode): Ditto.
(IPC::ArgumentCoder<WebCore::PaymentMerchantSession>::decode): Ditto.
(IPC::ArgumentCoder<WebCore::PaymentMethod>::encode): Ditto.
(IPC::ArgumentCoder<WebCore::PaymentMethod>::decode): Ditto.
* Shared/ios/InteractionInformationAtPosition.mm:
(WebKit::InteractionInformationAtPosition::encode const): Ditto.
(WebKit::InteractionInformationAtPosition::decode): Ditto.
* Shared/mac/WebCoreArgumentCodersMac.mm:
(IPC::ArgumentCoder<ProtectionSpace>::encodePlatformData): Ditto.
(IPC::ArgumentCoder<ProtectionSpace>::decodePlatformData): Ditto.
(IPC::ArgumentCoder<Credential>::encodePlatformData): Ditto.
(IPC::ArgumentCoder<Credential>::decodePlatformData): Ditto.
(IPC::ArgumentCoder<ContentFilterUnblockHandler>::encode): Ditto.
(IPC::ArgumentCoder<ContentFilterUnblockHandler>::decode): Ditto.
(IPC::ArgumentCoder<MediaPlaybackTargetContext>::encodePlatformData): Ditto.
(IPC::ArgumentCoder<MediaPlaybackTargetContext>::decodePlatformData): Ditto.
* Shared/mac/WebHitTestResultData.mm:
(WebKit::WebHitTestResultData::platformEncode const): Ditto.
(WebKit::WebHitTestResultData::platformDecode): Ditto.
* UIProcess/API/Cocoa/WKProcessPool.mm:
(-[WKProcessPool _setObject:forBundleParameter:]): Ditto.
(-[WKProcessPool _setObjectsForBundleParametersWithDictionary:]): Ditto.
* UIProcess/API/Cocoa/WKWebView.mm:
(-[WKWebView _setInputDelegate:]): Ditto.
* UIProcess/Cocoa/WebProcessPoolCocoa.mm:
(WebKit::WebProcessPool::platformInitializeWebProcess): Ditto.
* UIProcess/ios/PageClientImplIOS.mm:
(WebKit::PageClientImpl::startAssistingNode): Ditto.
* WebProcess/InjectedBundle/API/mac/WKWebProcessPlugInBrowserContextController.mm:
(-[WKWebProcessPlugInBrowserContextController _setFormDelegate:]): Ditto.
* WebProcess/InjectedBundle/mac/InjectedBundleMac.mm:
(WebKit::InjectedBundle::initialize): Ditto.
(WebKit::InjectedBundle::setBundleParameter): Ditto.
(WebKit::InjectedBundle::setBundleParameters): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225264 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWeb Inspector: Cleanup Inspector classes be more consistent about using fast malloc...
commit-queue@webkit.org [Wed, 29 Nov 2017 07:26:41 +0000 (07:26 +0000)]
Web Inspector: Cleanup Inspector classes be more consistent about using fast malloc / noncopyable
https://bugs.webkit.org/show_bug.cgi?id=180119

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2017-11-28
Reviewed by Devin Rousso.

Source/JavaScriptCore:

* inspector/InjectedScriptManager.h:
* inspector/JSGlobalObjectScriptDebugServer.h:
* inspector/agents/InspectorHeapAgent.h:
* inspector/agents/InspectorRuntimeAgent.h:
* inspector/agents/InspectorScriptProfilerAgent.h:
* inspector/agents/JSGlobalObjectRuntimeAgent.h:

Source/WebCore:

* inspector/InspectorCanvas.cpp:
* inspector/InspectorDatabaseResource.cpp:
* inspector/InspectorFrontendClientLocal.cpp:
* inspector/InspectorFrontendClientLocal.h:
* inspector/InspectorFrontendHost.cpp:
* inspector/InspectorHistory.h:
* inspector/InspectorInstrumentation.cpp:
* inspector/InspectorOverlay.cpp:
* inspector/InspectorShaderProgram.cpp:
* inspector/InspectorStyleSheet.cpp:
* inspector/InstrumentingAgents.cpp:
* inspector/PageScriptDebugServer.h:
* inspector/TimelineRecordFactory.cpp:
* inspector/WebInjectedScriptManager.cpp:
* inspector/WebInjectedScriptManager.h:
* inspector/WorkerScriptDebugServer.cpp:
* inspector/WorkerScriptDebugServer.h:
* inspector/agents/*
Be more consistent about namespace / fast malloc / noncopyable.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225263 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWebDriver: add an option to dump test results to a json file
carlosgc@webkit.org [Wed, 29 Nov 2017 07:25:53 +0000 (07:25 +0000)]
WebDriver: add an option to dump test results to a json file
https://bugs.webkit.org/show_bug.cgi?id=180082

Reviewed by Brian Burg.

Add --json-output command line option to run-webdriver-tests to dump test results to a json file in a format
compatible with the W3C report. WebDriverTestResult now represents a test file and contains a list of
subtests, instead of having one WebDriverTestResult per subtest. This way we can store also the harness result
and dump the results to different formats.

* Scripts/run-webdriver-tests:
* Scripts/webkitpy/webdriver_tests/webdriver_test_result.py:
(WebDriverTestResult.__init__):
(WebDriverTestResult):
(WebDriverTestResult.add_subtest_results):
* Scripts/webkitpy/webdriver_tests/webdriver_test_runner.py:
(WebDriverTestRunner.print_results):
(WebDriverTestRunner):
(WebDriverTestRunner.dump_results_to_json_file):
* Scripts/webkitpy/webdriver_tests/webdriver_test_runner_w3c.py:
(WebDriverTestRunnerW3C.run):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225262 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWeb Inspector: remove extra space before call frames in Canvas backtraces
webkit@devinrousso.com [Wed, 29 Nov 2017 07:09:15 +0000 (07:09 +0000)]
Web Inspector: remove extra space before call frames in Canvas backtraces
https://bugs.webkit.org/show_bug.cgi?id=180129

Reviewed by Matt Baker.

* UserInterface/Views/CanvasDetailsSidebarPanel.js:
(WI.CanvasDetailsSidebarPanel.prototype.initialLayout):
* UserInterface/Views/RecordingTraceDetailsSidebarPanel.js:
(WI.RecordingTraceDetailsSidebarPanel):
Hide the disclosure buttons of these TreeOutlines as they don't have children.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225261 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years ago[Cairo] Limit the number of active contexts in GraphicsContext3DCairo
zandobersek@gmail.com [Wed, 29 Nov 2017 07:07:35 +0000 (07:07 +0000)]
[Cairo] Limit the number of active contexts in GraphicsContext3DCairo
https://bugs.webkit.org/show_bug.cgi?id=166968

Reviewed by Alex Christensen.

Source/WebCore:

Cairo's implementation of GraphicsContext3D should follow the Mac's
and limit the number of active GraphicsContext3D objects.

The active contexts are listed in a global Deque, with the first
among them being recycled when the GraphicsContext3D::create()
function sees that the limit has been reached. That function still
returns null if even after recycling the number of contexts didn't
decrease.

Finally, in the GraphicsContext3D destructor, the context being
destroyed is removed from the list of active contexts.

No new tests -- relevant tests are now passing.

* platform/graphics/cairo/GraphicsContext3DCairo.cpp:
(WebCore::activeContexts):
(WebCore::GraphicsContext3D::create):
(WebCore::GraphicsContext3D::~GraphicsContext3D):

LayoutTests:

* platform/gtk/TestExpectations: Unskip WebGL tests that were failing
due to missing active context limit management.
* platform/wpe/TestExpectations: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225260 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWeb Inspector: Clean up backtrace in Canvas details sidebar
mattbaker@apple.com [Wed, 29 Nov 2017 06:16:54 +0000 (06:16 +0000)]
Web Inspector: Clean up backtrace in Canvas details sidebar
https://bugs.webkit.org/show_bug.cgi?id=179807
<rdar://problem/35604378>

Reviewed by Devin Rousso.

* UserInterface/Controllers/CallFrameTreeController.js: Added.
Display a list of call frames in a tree outline. Browse to the represented
call frame's source code location when a tree element is clicked or selected.

(WI.CallFrameTreeController):
(WI.CallFrameTreeController.prototype.get treeOutline):
(WI.CallFrameTreeController.prototype.get callFrames):
(WI.CallFrameTreeController.prototype.set callFrames):
(WI.CallFrameTreeController.prototype.disconnect):
(WI.CallFrameTreeController.prototype._treeElementClicked):
(WI.CallFrameTreeController.prototype._treeSelectionDidChange):
(WI.CallFrameTreeController.prototype._showSourceCodeLocation):

* UserInterface/Main.html:

* UserInterface/Views/CanvasDetailsSidebarPanel.css:
(.sidebar > .panel.details.canvas .details-section.canvas-backtrace .call-frame): Deleted.
* UserInterface/Views/CanvasDetailsSidebarPanel.js:
(WI.CanvasDetailsSidebarPanel.prototype.initialLayout):
(WI.CanvasDetailsSidebarPanel.prototype._refreshBacktraceSection):

* UserInterface/Views/RecordingTraceDetailsSidebarPanel.css:
(.sidebar > .panel.details.recording-trace > .content > .call-frame): Deleted.
* UserInterface/Views/RecordingTraceDetailsSidebarPanel.js:
(WI.RecordingTraceDetailsSidebarPanel):
(WI.RecordingTraceDetailsSidebarPanel.prototype.updateAction):

* UserInterface/Views/TreeElement.js:
(WI.TreeElement.treeElementToggled):
(WI.TreeElement.prototype.selectOnMouseDown):
Prevent selection if parent tree outline is not selectable.

* UserInterface/Views/TreeOutline.css:
(.tree-outline.non-selectable .item:hover):

* UserInterface/Views/TreeOutline.js:
Add `selectable` behavior, set at construction time. When false,
clicking a tree element dispatches an event instead of selecting
the tree element. Default true.

(WI.TreeOutline):
(WI.TreeOutline.prototype.get selectable):
Dispatch click event when not selectable, and some drive-by cleanup.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225259 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWeb Inspector: Write script syntax tree tests for template literals and default param...
commit-queue@webkit.org [Wed, 29 Nov 2017 06:10:50 +0000 (06:10 +0000)]
Web Inspector: Write script syntax tree tests for template literals and default parameter values
https://bugs.webkit.org/show_bug.cgi?id=149450
<rdar://problem/22796879>

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2017-11-28
Reviewed by Devin Rousso.

* inspector/model/parse-script-syntax-tree-expected.txt:
* inspector/model/parse-script-syntax-tree.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225258 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWeb Inspector: Move console Preserve Log setting from Setting tab to Console navigati...
commit-queue@webkit.org [Wed, 29 Nov 2017 05:08:21 +0000 (05:08 +0000)]
Web Inspector: Move console Preserve Log setting from Setting tab to Console navigation bar
https://bugs.webkit.org/show_bug.cgi?id=180125

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2017-11-28
Reviewed by Matt Baker.

* Localizations/en.lproj/localizedStrings.js:
* UserInterface/Views/LogContentView.js:
(WI.LogContentView):
(WI.LogContentView.prototype.get navigationItems):
(WI.LogContentView.prototype._clearLogOnNavigateSettingChanged):
* UserInterface/Views/SettingsTabContentView.js:
(WI.SettingsTabContentView.prototype._createGeneralSettingsView):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225257 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWeb Inspector: Remove Network "Clear on load" from Settings tab now that Network...
commit-queue@webkit.org [Wed, 29 Nov 2017 04:44:50 +0000 (04:44 +0000)]
Web Inspector: Remove Network "Clear on load" from Settings tab now that Network tab has a toggle for it
https://bugs.webkit.org/show_bug.cgi?id=180123

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2017-11-28
Reviewed by Matt Baker.

* Localizations/en.lproj/localizedStrings.js:
* UserInterface/Views/SettingsTabContentView.js:
(WI.SettingsTabContentView.prototype._createGeneralSettingsView):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225256 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoClean up spanners before creating nested column context
zalan@apple.com [Wed, 29 Nov 2017 04:12:32 +0000 (04:12 +0000)]
Clean up spanners before creating nested column context
https://bugs.webkit.org/show_bug.cgi?id=180107
<rdar://problem/35686655>

Reviewed by Antti Koivisto.

Source/WebCore:

When an existing spanner placeholder is moved into a newly constructed (and nested)
multicolumn context, we figure it's not valid anymore and end up destroying it
(see RenderMultiColumnFlow::fragmentedFlowDescendantInserted).
This is very unfortunate since as we climb back on the stack, we could hit this renderer as
the newly inserted child.

This patch proactively removes the invalid placeholders and moves the associated spanners back to their
original position.

Test: fast/multicol/crash-when-constructing-nested-columns.html

* rendering/RenderMultiColumnFlow.h:
* style/RenderTreeUpdaterMultiColumn.cpp:
(WebCore::RenderTreeUpdater::MultiColumn::createFragmentedFlow):
RenderTreeUpdater::MultiColumn::destroyFragmentedFlow still relies on the placeholder removal
logic in RenderMultiColumnFlow::fragmentedFlowDescendantInserted.

LayoutTests:

* fast/multicol/crash-when-constructing-nested-columns-expected.txt: Added.
* fast/multicol/crash-when-constructing-nested-columns.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225255 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoREGRESSION (High Sierra): Layout Test fast/multicol/newmulticol/spanner2.html is...
timothy_horton@apple.com [Wed, 29 Nov 2017 02:59:47 +0000 (02:59 +0000)]
REGRESSION (High Sierra): Layout Test fast/multicol/newmulticol/spanner2.html is a flaky image failure on WK1
https://bugs.webkit.org/show_bug.cgi?id=177826
<rdar://problem/34876498>

Reviewed by Alexey Proskuryakov.

Source/WebCore:

No new tests, fixes some already-broken tests.

* platform/graphics/mac/PDFDocumentImageMac.mm:
(WebCore::PDFDocumentImage::drawPDFPage):
Save and restore one more piece of context state that
PDFKit mutates while drawing.

Source/WebCore/PAL:

* pal/spi/cg/CoreGraphicsSPI.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225254 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoNetworkCache::Storage should protect itself when removing operations from its maps
commit-queue@webkit.org [Wed, 29 Nov 2017 02:23:38 +0000 (02:23 +0000)]
NetworkCache::Storage should protect itself when removing operations from its maps
https://bugs.webkit.org/show_bug.cgi?id=180118

Patch by Youenn Fablet <youenn@apple.com> on 2017-11-28
Reviewed by Antti Koivisto.

The operations can contain ref to the Storage object and removing them from the map may destroy the Storage object

* NetworkProcess/cache/NetworkCacheStorage.cpp:
(WebKit::NetworkCache::Storage::remove):
(WebKit::NetworkCache::Storage::finishReadOperation):
(WebKit::NetworkCache::Storage::finishWriteOperation):
(WebKit::NetworkCache::Storage::traverse):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225253 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoAllow attachment elements with no appearance to defer rendering to child nodes
wenson_hsieh@apple.com [Wed, 29 Nov 2017 02:06:22 +0000 (02:06 +0000)]
Allow attachment elements with no appearance to defer rendering to child nodes
https://bugs.webkit.org/show_bug.cgi?id=180117
<rdar://problem/35735339>

Reviewed by Tim Horton.

Source/WebCore:

Test: fast/attachment/attachment-without-appearance.html

When -webkit-appearance: none; is specified on an attachment element, allow it to fall back to rendering its
subtree. Currently, attachment elements without an appearance don't allow this and truncate the render tree at
the RenderAttachment, since RenderAttachment cannot have any children.

In a followup, this will enable us to render a shadow subtree under the attachment element to display in-place
attachment content, and easily toggle between in-place and icon display by changing the appearance.

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

If no appearance is specified, emit a RenderBlockFlow instead of a RenderAttachment.

(WebCore::HTMLAttachmentElement::setFile):
(WebCore::HTMLAttachmentElement::attachmentRenderer const):

Renamed from renderer(). HTMLAttachmentElement::renderer() now uses the superclass' implementation, and no
longer returns a RenderAttachment in all circumstances. Instead, places that expect a RenderAttachment now go
through HTMLAttachmentElement::renderAttachment() instead.

(WebCore::HTMLAttachmentElement::parseAttribute):
* html/HTMLAttachmentElement.h:
* page/DragController.cpp:
(WebCore::DragController::startDrag):
* rendering/RenderAttachment.h:
(WebCore::HTMLAttachmentElement::renderer const): Deleted.

LayoutTests:

Adds a ref test verifying that an appearance-less attachment can render child nodes. More extensive testing to
come in a followup patch.

* fast/attachment/attachment-without-appearance-expected.html: Added.
* fast/attachment/attachment-without-appearance.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225252 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoRegister Documents as ServiceWorker clients to the StorageProcess
commit-queue@webkit.org [Wed, 29 Nov 2017 01:50:03 +0000 (01:50 +0000)]
Register Documents as ServiceWorker clients to the StorageProcess
https://bugs.webkit.org/show_bug.cgi?id=180047

Patch by Youenn Fablet <youenn@apple.com> on 2017-11-28
Reviewed by Brady Eidson.

Source/WebCore:

No change of behavior.
These changes will be covered when implementing ServiceWorker Clients API.

Registering a document when being created by DocumentLoader.
In the future, we may restrict registration to only documents related to origins that have some ongoing service worker activity.
This would require to keep track of which documents are registered so that we unregister these ones.
This would also require to register existing documents when one of the document with the same origin starts registering a service worker.
Unregistering a document inside prepareForDestruction.

Storing all clients in SWServer as a HasMap keyed by ClientOrigin.
Processing will then iterate through the list of clients scoped by the client origin.

Adding a ClientOrigin class as a pair of top origin and frame origin since service workers will be related based on that information.

* WebCore.xcodeproj/project.pbxproj:
* dom/Document.cpp:
(WebCore::Document::prepareForDestruction):
* loader/DocumentLoader.cpp:
(WebCore::DocumentLoader::commitData):
* page/ClientOrigin.h: Added.
(WebCore::ClientOrigin::emptyKey):
(WebCore::ClientOrigin::hash const):
(WebCore::ClientOrigin::operator== const):
(WTF::ClientOriginKeyHash::hash):
(WTF::ClientOriginKeyHash::equal):
(WTF::HashTraits<WebCore::ClientOrigin>::emptyValue):
(WTF::HashTraits<WebCore::ClientOrigin>::constructDeletedValue):
(WTF::HashTraits<WebCore::ClientOrigin>::isDeletedValue):
* workers/service/ServiceWorkerClientIdentifier.h:
(WebCore::ServiceWorkerClientIdentifier::operator== const):
* workers/service/server/SWClientConnection.h:
* workers/service/server/SWServer.cpp:
(WebCore::SWServer::registerServiceWorkerClient):
(WebCore::SWServer::unregisterServiceWorkerClient):
* workers/service/server/SWServer.h:

Source/WebKit:

Adding IPC to register/unregister clients to the storage process.
This will be used to enable service worker termination, and implementation of Clients API.

WebSWServerConnection keeps a hash map of all its related clients.
This allows unregistering these clients if the corresponding web process crashes.

* StorageProcess/ServiceWorker/WebSWServerConnection.cpp:
(WebKit::WebSWServerConnection::~WebSWServerConnection):
(WebKit::WebSWServerConnection::registerServiceWorkerClient):
(WebKit::WebSWServerConnection::unregisterServiceWorkerClient):
* StorageProcess/ServiceWorker/WebSWServerConnection.h:
* StorageProcess/ServiceWorker/WebSWServerConnection.messages.in:
* StorageProcess/ServiceWorker/WebSWServerToContextConnection.h:
* WebProcess/Storage/WebSWClientConnection.cpp:
(WebKit::WebSWClientConnection::registerServiceWorkerClient):
(WebKit::WebSWClientConnection::unregisterServiceWorkerClient):
* WebProcess/Storage/WebSWClientConnection.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225251 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWeb Inspector: Network Tab - Add a toggle in the network tab to control automatically...
commit-queue@webkit.org [Wed, 29 Nov 2017 01:47:01 +0000 (01:47 +0000)]
Web Inspector: Network Tab - Add a toggle in the network tab to control automatically clearing or preserving log across loads
https://bugs.webkit.org/show_bug.cgi?id=180110
<rdar://problem/34071789>

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2017-11-28
Reviewed by Timothy Hatcher.

* Localizations/en.lproj/localizedStrings.js:
New strings.

* UserInterface/Views/CheckboxNavigationItem.css:
(.navigation-bar .item.checkbox input[type=checkbox]):
Tweak style to more center the checkbox vertically.

* UserInterface/Views/CheckboxNavigationItem.js:
(WI.CheckboxNavigationItem):
Fix setting the initial value of a checkbox navigation item.

* UserInterface/Views/NetworkTableContentView.js:
(WI.NetworkTableContentView):
(WI.NetworkTableContentView.prototype.get navigationItems):
(WI.NetworkTableContentView.prototype.closed):
(WI.NetworkTableContentView.prototype._clearNetworkOnNavigateSettingChanged):
Add a new checkbox for the clear on navigation setting.

* UserInterface/Views/SettingEditor.js:
(WI.SettingEditor.createForSetting):
Update the editor checkbox if the setting changes outside of the setting editor.

* UserInterface/Views/NavigationItem.js:
(WI.NavigationItem.prototype.get tooltip):
(WI.NavigationItem.prototype.set tooltip):
* UserInterface/Views/ActivateButtonNavigationItem.js:
(WI.ActivateButtonNavigationItem.prototype.set activated):
* UserInterface/Views/ButtonNavigationItem.js:
(WI.ButtonNavigationItem):
(WI.ButtonNavigationItem.prototype.get toolTip): Deleted.
(WI.ButtonNavigationItem.prototype.set toolTip): Deleted.
* UserInterface/Views/ToggleButtonNavigationItem.js:
(WI.ToggleButtonNavigationItem.prototype.set alternateToolTip):
(WI.ToggleButtonNavigationItem.prototype.set toggled):
Move tooltip to the base class and rename it from `toolTip` to `tooltip` to
match existing generic places like TreeElement.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225250 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoServiceWorkerGlobalScope.clients should always return the same object
cdumez@apple.com [Wed, 29 Nov 2017 01:33:29 +0000 (01:33 +0000)]
ServiceWorkerGlobalScope.clients should always return the same object
https://bugs.webkit.org/show_bug.cgi?id=180116

Reviewed by Geoffrey Garen.

Source/WebCore:

ServiceWorkerGlobalScope.clients should always return the same object, as per the specification:
- https://w3c.github.io/ServiceWorker/#serviceworkerglobalscope-interface

Test: http/tests/workers/service/ServiceWorkerGlobalScope_clients_SameObject.html

* Sources.txt:
* WebCore.xcodeproj/project.pbxproj:
* bindings/js/JSServiceWorkerGlobalScopeCustom.cpp: Added.
(WebCore::JSServiceWorkerGlobalScope::visitAdditionalChildren):
* workers/service/ServiceWorkerClients.idl:
* workers/service/ServiceWorkerGlobalScope.idl:

LayoutTests:

Add layout test coverage.

* http/tests/workers/service/ServiceWorkerGlobalScope_clients_SameObject-expected.txt: Added.
* http/tests/workers/service/ServiceWorkerGlobalScope_clients_SameObject.html: Added.
* http/tests/workers/service/resources/ServiceWorkerGlobalScope_clients_SameObject-worker.js: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225249 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoGet rid of ServiceWorker::allWorkers() hashmap
cdumez@apple.com [Wed, 29 Nov 2017 00:28:42 +0000 (00:28 +0000)]
Get rid of ServiceWorker::allWorkers() hashmap
https://bugs.webkit.org/show_bug.cgi?id=180111

Reviewed by Brady Eidson.

Get rid of ServiceWorker::allWorkers() hashmap as it is not thread safe and we'll soon have
ServiceWorker objects living in various service worker threads.

Instead, we now have a per-ScriptExecutionContext map, which is inherently thread-safe.

No new tests, no web-facing behavior change.

* dom/ScriptExecutionContext.cpp:
(WebCore::ScriptExecutionContext::registerServiceWorker):
(WebCore::ScriptExecutionContext::unregisterServiceWorker):
* dom/ScriptExecutionContext.h:
(WebCore::ScriptExecutionContext::serviceWorker):
* workers/service/ServiceWorker.cpp:
(WebCore::ServiceWorker::getOrCreate):
(WebCore::ServiceWorker::ServiceWorker):
(WebCore::ServiceWorker::~ServiceWorker):
(WebCore::ServiceWorker::stop):
* workers/service/ServiceWorker.h:
* workers/service/server/SWClientConnection.cpp:
(WebCore::SWClientConnection::updateWorkerState):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225248 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoUnreviewed, correct the location for an expected result file.
ryanhaddad@apple.com [Wed, 29 Nov 2017 00:23:29 +0000 (00:23 +0000)]
Unreviewed, correct the location for an expected result file.

* platform/mac-elcapitan/fast/forms/alternative-presentation-button/replacement-expected.txt: Renamed from LayoutTests/platform/mac-elcapitan/fast/alternative-presentation-button/replacement-expected.txt.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225247 268f45cc-cd09-0410-ab3c-d52691b4dbfc

3 years agoWeb Inspector: Include Beacon loads in the Network Table's "Other" filter
commit-queue@webkit.org [Wed, 29 Nov 2017 00:01:54 +0000 (00:01 +0000)]
Web Inspector: Include Beacon loads in the Network Table's "Other" filter
https://bugs.webkit.org/show_bug.cgi?id=180113

Patch by Joseph Pecoraro <pecoraro@apple.com> on 2017-11-28
Reviewed by Matt Baker.

* UserInterface/Views/NetworkTableContentView.js:
(WI.NetworkTableContentView):
Ensure the Other filter will handle any type that hasn't already been handled.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@225246 268f45cc-cd09-0410-ab3c-d52691b4dbfc