Source/WTF/ChangeLog

2016-10-12  Ryan Haddad  <ryanhaddad@apple.com>

        Unreviewed, rolling out r207225.

        This change causes debug tests to exit early with crashes.

        Reverted changeset:

        "Optional's move-constructor and move-assignment operator
        don't disengage the value being moved from"
        https://bugs.webkit.org/show_bug.cgi?id=163309
        http://trac.webkit.org/changeset/207225

2016-10-11  Sam Weinig  <sam@webkit.org>

        Optional's move-constructor and move-assignment operator don't disengage the value being moved from
        https://bugs.webkit.org/show_bug.cgi?id=163309

        Reviewed by Anders Carlsson.

        * wtf/Optional.h:
        (WTF::Optional::Optional):
        (WTF::Optional::operator=):
        Disengage 'other' on move-construction and move-assignment.

2016-10-08  Filip Pizlo  <fpizlo@apple.com>

        MarkedBlock should know what objects are live during marking
        https://bugs.webkit.org/show_bug.cgi?id=162309

        Reviewed by Geoffrey Garen.
        
        This removes the atomicity mode, because it's not really used: it only affects the
        concurrentBlah methods, but their only users turn on atomicity. This was useful because
        previously, some binary Bitmap methods (like merge(const Bitmap&)) couldn't be used
        effectively in the GC because some of the GC's bitmaps set the atomic mode and some didn't.
        Removing this useless mode is the best solution.
        
        Also added some new binary Bitmap methods: mergeAndClear(Bitmap& other) and
        setAndClear(Bitmap& other). They perform their action on 'this' (either merge or set,
        respectively) while also clearing the contents of 'other'. This is great for one of the GC
        hot paths.

        * wtf/Bitmap.h:
        (WTF::WordType>::Bitmap):
        (WTF::WordType>::get):
        (WTF::WordType>::set):
        (WTF::WordType>::testAndSet):
        (WTF::WordType>::testAndClear):
        (WTF::WordType>::concurrentTestAndSet):
        (WTF::WordType>::concurrentTestAndClear):
        (WTF::WordType>::clear):
        (WTF::WordType>::clearAll):
        (WTF::WordType>::nextPossiblyUnset):
        (WTF::WordType>::findRunOfZeros):
        (WTF::WordType>::count):
        (WTF::WordType>::isEmpty):
        (WTF::WordType>::isFull):
        (WTF::WordType>::merge):
        (WTF::WordType>::filter):
        (WTF::WordType>::exclude):
        (WTF::WordType>::forEachSetBit):
        (WTF::WordType>::mergeAndClear):
        (WTF::WordType>::setAndClear):
        (WTF::=):
        (WTF::WordType>::hash):

2016-10-11  Said Abou-Hallawa  <sabouhallawa@apple.com>

        Add SynchronizedFixedQueue class
        https://bugs.webkit.org/show_bug.cgi?id=162478

        Reviewed by Geoffrey Garen.

        This class represents a simple producer/consumer worker. It facilitates
        synchronizing enqueuing to and dequeuing from a fixed size-queue. It uses
        a single lock and a single condition to synchronize all its members among
        the working threads. This means a single thread is active at any time and
        and the other threads are blocked waiting for the lock to be released. Or
        they are sleeping waiting for the condition to be satisfied.

        * WTF.xcodeproj/project.pbxproj:
        * wtf/SynchronizedFixedQueue.h: Added.
        (WTF::SynchronizedFixedQueue::SynchronizedFixedQueue):
        (WTF::SynchronizedFixedQueue::open): Restore the queue to its original state.
        (WTF::SynchronizedFixedQueue::close): Wake all the sleeping threads with a closing state.
        (WTF::SynchronizedFixedQueue::isOpen): Does the queue accept new items?
        (WTF::SynchronizedFixedQueue::enqueue): Enqueue an item into the queue.
        (WTF::SynchronizedFixedQueue::dequeue): Dequeue an item form the queue.

2016-10-11  Alex Christensen  <achristensen@webkit.org>

        Remove dead networking code
        https://bugs.webkit.org/show_bug.cgi?id=163263

        Reviewed by Daniel Bates.

        * wtf/Platform.h:
        * wtf/SchedulePair.h:
        * wtf/SchedulePairMac.mm:

2016-10-10  Zan Dobersek  <zdobersek@igalia.com>

        Add ENABLE_ENCRYPTED_MEDIA configuration option
        https://bugs.webkit.org/show_bug.cgi?id=163219

        Reviewed by Darin Adler.

        * wtf/FeatureDefines.h:
        If undefined, define the ENABLE_ENCRYPTED_MEDIA option to 0.

2016-10-10  Yusuke Suzuki  <utatane.tea@gmail.com>

        [DOMJIT] Implement Node accessors in DOMJIT
        https://bugs.webkit.org/show_bug.cgi?id=163005

        Reviewed by Filip Pizlo.

        Add CAST_OFFSET. It is not necessary for JSCell thingy
        since we don't use virtual member functions. However, it
        is not true for WebCore DOM wrapped objects.

        * wtf/StdLibExtras.h:

2016-10-07  Chris Dumez  <cdumez@apple.com>

        window.navigator.language incorrectly returns all lowercase string
        https://bugs.webkit.org/show_bug.cgi?id=163096

        Reviewed by Darin Adler.

        Update platformUserPreferredLanguages() so that it no longer lowercases
        the string it returns. On Mac, we rely on CFLocale which returns
        BCP-47 language tags as per:
        - https://developer.apple.com/reference/corefoundation/1666963-cflocale?language=objc

        * wtf/PlatformUserPreferredLanguagesMac.mm:
        (WTF::httpStyleLanguageCode):
        * wtf/PlatformUserPreferredLanguagesUnix.cpp:
        (WTF::platformLanguage):

2016-10-06  Brent Fulgham  <bfulgham@apple.com>

        [Win][Direct2D] Add Direct2D CMake rules
        https://bugs.webkit.org/show_bug.cgi?id=162925

        Reviewed by Brent Fulgham.

        * wtf/Platform.h: Don't USE(CA) or USE(CG) if building
        with Direct2D.

2016-10-05  Yusuke Suzuki  <utatane.tea@gmail.com>

        [DOMJIT] Add initial CheckDOM and CallDOM implementations
        https://bugs.webkit.org/show_bug.cgi?id=162941

        Reviewed by Filip Pizlo.

        * wtf/Box.h:
        (WTF::Box::Box):

2016-10-05  Zan Dobersek  <zdobersek@igalia.com>

        Rename ENABLE_ENCRYPTED_MEDIA_V2 to ENABLE_LEGACY_ENCRYPTED_MEDIA
        https://bugs.webkit.org/show_bug.cgi?id=162903

        Reviewed by Alex Christensen.

        Rename build guards for the remaining implementation of the legacy EME API
        to ENABLE_LEGACY_ENCRYPTED_MEDIA. This will allow for the future implementation
        of the near-finished API to be guarded with the simple ENABLE_ENCRYPTED_MEDIA guards.

        * wtf/FeatureDefines.h:

2016-10-04  Saam Barati  <sbarati@apple.com>

        String.prototype.toLowerCase should be a DFG/FTL intrinsic
        https://bugs.webkit.org/show_bug.cgi?id=162887

        Reviewed by Filip Pizlo and Yusuke Suzuki.

        This patch exposes a new StringImpl function called convertToLowercaseWithoutLocaleStartingAtFailingIndex8Bit
        which extracts slow path for the 8-bit part of convertToLowercaseWithoutLocale
        into a helper function. I decided to extract this into its own function because
        it may be the case that JSCs JITs will want to continue the operation
        after it has already ensured that part of an 8-bit string is lower case.

        * wtf/text/StringImpl.cpp:
        (WTF::StringImpl::convertToLowercaseWithoutLocale):
        (WTF::StringImpl::convertToLowercaseWithoutLocaleStartingAtFailingIndex8Bit):
        * wtf/text/StringImpl.h:
        * wtf/text/WTFString.cpp:
        (WTF::String::convertToLowercaseWithoutLocaleStartingAtFailingIndex8Bit):
        * wtf/text/WTFString.h:

2016-10-04  Chris Dumez  <cdumez@apple.com>

        Implement KeyboardEvent.code from the UI Event spec
        https://bugs.webkit.org/show_bug.cgi?id=149584

        Reviewed by Darin Adler.

        Add build time flag to toggle support for the code attribute on
        KeyboardEvent and only enable it on Mac for now, given that the
        implementation is missing on other platforms.

        * wtf/FeatureDefines.h:

2016-10-03  Chris Dumez  <cdumez@apple.com>

        Add support for KeyboardEvent.key attribute
        https://bugs.webkit.org/show_bug.cgi?id=36267

        Reviewed by Darin Adler.

        Add compile time flag for the key attribute on KeyboardEvent and enable
        it on Cocoa only.

        * wtf/FeatureDefines.h:

2016-09-29  Sam Weinig  <sam@webkit.org>

        Add initial support for IDL union conversion
        https://bugs.webkit.org/show_bug.cgi?id=161576

        Reviewed by Chris Dumez.

        * WTF.xcodeproj/project.pbxproj:
        * wtf/CMakeLists.txt:
        Add Brigand.h

        * wtf/Brigand.h: Added.
        Import a standalone copy of Edouard Alligand and Joel Falcou's 
        Brigand library for help with for help with list based meta programming

        * wtf/StdLibExtras.h:
        Add a new three new type traits, IsTemplate, IsBaseOfTemplate and RemoveCVAndReference.
        - IsTemplate acts like std::is_class, but works with a generic base.
        - IsBaseOfTemplate acts like std::is_base_of, but works with a generic base.
        - RemoveCVAndReference combines std::remove_cv and std::remove_reference.

2016-09-30  Filip Pizlo  <fpizlo@apple.com>

        B3::moveConstants should be able to edit code to minimize the number of constants
        https://bugs.webkit.org/show_bug.cgi?id=162764

        Reviewed by Saam Barati.
        
        I thought it would be a good idea to document the fact that dominator traversal happens in a
        particular order for a reason.

        * wtf/Dominators.h:

2016-09-29  Filip Pizlo  <fpizlo@apple.com>

        Air should have a way of expressing additional instruction flags
        https://bugs.webkit.org/show_bug.cgi?id=162699

        Reviewed by Mark Lam.

        * wtf/CommaPrinter.h:
        (WTF::CommaPrinter::CommaPrinter):
        (WTF::CommaPrinter::dump):
        (WTF::CommaPrinter::didPrint):

2016-09-30  Youenn Fablet  <youenn@apple.com>

        Add a way to go from a RefPtr<T> to Ref<const T>
        https://bugs.webkit.org/show_bug.cgi?id=162683

        Reviewed by Alex Christensen.

        * wtf/RefPtr.h:
        (WTF::RefPtr::releaseConstNonNull): Added.

2016-09-29  Mark Lam  <mark.lam@apple.com>

        Re-enable StringView life-cycle checking.
        https://bugs.webkit.org/show_bug.cgi?id=160384
        <rdar://problem/28479434>

        Reviewed by Saam Barati.

        Re-landing after slow running tests have been resolved.

        * wtf/text/StringView.h:

2016-09-29  Commit Queue  <commit-queue@webkit.org>

        Unreviewed, rolling out r206563.
        https://bugs.webkit.org/show_bug.cgi?id=162732

        Caused stress/op_*.js.ftl-no-cjit tests to time out (Requested
        by ryanhaddad on #webkit).

        Reverted changeset:

        "Re-enable StringView life-cycle checking."
        https://bugs.webkit.org/show_bug.cgi?id=160384
        http://trac.webkit.org/changeset/206563

2016-09-29  Fujii Hironori  <Hironori.Fujii@sony.com>

        Clang 3.9 reports a compilation warning about ENABLE_EXCEPTION_SCOPE_VERIFICATION
        https://bugs.webkit.org/show_bug.cgi?id=162718

        Reviewed by Alex Christensen.

        Clang 3.9 reports a following compilation warning:
          Source/JavaScriptCore/runtime/VM.h:656:5: warning: macro expansion producing 'defined' has undefined behavior [-Wexpansion-to-defined]

        * wtf/Platform.h: Changed the definition of ENABLE_EXCEPTION_SCOPE_VERIFICATION not to use 'defined'.

2016-09-28  Mark Lam  <mark.lam@apple.com>

        Re-enable StringView life-cycle checking.
        https://bugs.webkit.org/show_bug.cgi?id=160384
        <rdar://problem/28479434>

        Reviewed by Saam Barati.

        * wtf/text/StringView.h:

2016-09-28  Filip Pizlo  <fpizlo@apple.com>

        The write barrier should be down with TSO
        https://bugs.webkit.org/show_bug.cgi?id=162316

        Reviewed by Geoffrey Garen.
        
        Added clearRange(), which quickly clears a range of bits. This turned out to be useful for
        a DFG optimization pass.

        * wtf/FastBitVector.cpp:
        (WTF::FastBitVector::clearRange):
        * wtf/FastBitVector.h:

2016-09-28  Mark Lam  <mark.lam@apple.com>

        Fix race condition in StringView's UnderlyingString lifecycle management.
        https://bugs.webkit.org/show_bug.cgi?id=162702

        Reviewed by Geoffrey Garen.

        There 2 relevant functions at play:

        void StringView::setUnderlyingString(const StringImpl* string)
        {
            UnderlyingString* underlyingString;
            if (!string)
                underlyingString = nullptr;
            else {
                std::lock_guard<StaticLock> lock(underlyingStringsMutex);
                auto result = underlyingStrings().add(string, nullptr);
                if (result.isNewEntry)
                    result.iterator->value = new UnderlyingString(*string);
                else
                    ++result.iterator->value->refCount;
                underlyingString = result.iterator->value; // Point P2.
            }
            adoptUnderlyingString(underlyingString); // Point P5.
        }

        ... and ...

        void StringView::adoptUnderlyingString(UnderlyingString* underlyingString)
        {
            if (m_underlyingString) {
                // Point P0.
                if (!--m_underlyingString->refCount) {
                    if (m_underlyingString->isValid) { // Point P1.
                        std::lock_guard<StaticLock> lock(underlyingStringsMutex);
                        underlyingStrings().remove(&m_underlyingString->string); // Point P3.
                    }
                    delete m_underlyingString; // Point P4.
                }
            }
            m_underlyingString = underlyingString;
        }

        Imagine the following scenario:

        1. Thread T1 has been using an UnderlyingString U1, and is now done with it.
           T1 runs up to point P1 in adoptUnderlyingString(), and is blocked waiting for
           the underlyingStringsMutex (which is currently being held by Thread T2).
        2. Context switch to Thread T2.
           T2 wants to use UnderlyingString U1, and runs up to point P2 in setUnderlyingString()
           and releases the underlyingStringsMutex.
           Note: T2 thinks it has successfully refCounted U1, and therefore U1 is safe to use.
        3. Context switch to Thread T1.
           T1 acquires the underlyingStringsMutex, and proceeds to remove it from the
           underlyingStrings() map (see Point P3).  It thinks it has successfully done so
           and proceeds to delete U1 (see Point P4).
        4. Context switch to Thread T2.
           T2 proceeds to use U1 (see Point P5 in setUnderlyingString()).
           Note: U1 has already been freed.  This is a use after free.

        The fix is to acquire the underlyingStringsMutex at Point P0 in adoptUnderlyingString()
        instead of after P1.  This ensures that the decrementing of the UnderlyingString
        refCount and its removal from the underlyingStrings() map is done as an atomic unit.

        Note: If you look in StringView.cpp, you see another setUnderlyingString() which
        takes a StringView otherString.  This version of setUnderlyingString() can only
        be called from within the same thread that created the other StringView.  As a
        result, here, we are guaranteed that the UnderlyingString refCount is never zero,
        and there's no other threat of another thread trying to delete the UnderlyingString
        while we adopt it.  Hence, we don't need to acquire the underlyingStringsMutex
        here.

        This race condition was found when running layout tests fetch/fetch-worker-crash.html
        and storage/indexeddb/modern/opendatabase-versions.html when CHECK_STRINGVIEW_LIFETIME
        is enabled.  This issue resulted in those tests crashing due to a use-after-free.

        * wtf/text/StringView.cpp:
        (WTF::StringView::adoptUnderlyingString):
        (WTF::StringView::setUnderlyingString):

2016-09-28  Brent Fulgham  <bfulgham@apple.com>

        Correct 'safeCast' implementation
        https://bugs.webkit.org/show_bug.cgi?id=162679
        <rdar://problem/28518189>

        Reviewed by Zalan Bujtas.

        * wtf/StdLibExtras.h:
        (WTF::safeCast): Use a RELEASE_ASSERT.

2016-09-27  Don Olmstead  <don.olmstead@am.sony.com>

        [CMake] Add HAVE_LOCALTIME_R definition
        https://bugs.webkit.org/show_bug.cgi?id=162636

        Reviewed by Alex Christensen.

        * wtf/DateMath.cpp:
        (WTF::getLocalTime):
        * wtf/GregorianDateTime.cpp:
        (WTF::GregorianDateTime::setToCurrentLocalTime):
        * wtf/Platform.h:

2016-09-27  JF Bastien  <jfbastien@apple.com>

        Speed up Heap::isMarkedConcurrently
        https://bugs.webkit.org/show_bug.cgi?id=162095

        Reviewed by Filip Pizlo.

        Heap::isMarkedConcurrently had a load-load fence which is expensive on weak memory ISAs such as ARM.

        This patch is fairly perf-neutral overall, but the GC's instrumentation reports:
          GC:Eden is 93% average runtime after change
          GC:Full is 76% average runtime after change

        The fence was there because:
         1. If the read of m_markingVersion in MarkedBlock::areMarksStale isn't what's expected then;
         2. The read of m_marks in MarkedBlock::isMarked needs to observe the value that was stored *before* m_markingVersion was stored.

        This ordering isn't guaranteed on ARM, which has a weak memory model.

        There are 3 ways to guarantee this ordering:
         A. Use a barrier instruction.
         B. Use a load-acquire (new in ARMv8).
         C. use ARM's address dependency rule, which C++ calls memory_order_consume.

        In general:
         A. is slow but orders all of memory in an intuitive manner.
         B. is faster-ish and has the same property-ish.
         C. should be faster still, but *only orders dependent loads*. This last part is critical! Consume isn't an all-out replacement for acquire (acquire is rather a superset of consume).

        ARM explains the address dependency rule in their document "barrier litmus tests and cookbook":

        > *Resolving by the use of barriers and address dependency*
        >
        > There is a rule within the ARM architecture that:
        > Where the value returned by a read is used to compute the virtual address of a subsequent read or write (this is known as an address dependency), then these two memory accesses will be observed in program order. An address dependency exists even if the value read by the first read has no effect in changing the virtual address (as might be the case if the value returned is masked off before it is used, or if it had no effect on changing a predicted address value).
        > This restriction applies only when the data value returned from one read is used as a data value to calculate the address of a subsequent read or write. This does not apply if the data value returned from one read is used to determine the condition code flags, and the values of the flags are used for condition code evaluation to determine the address of a subsequent reads, either through conditional execution or the evaluation of a branch. This is known as a control dependency.
        > Where both a control and address dependency exist, the ordering behaviour is consistent with the address dependency.

        C++'s memory_order_consume is unfortunately unimplemented by C++ compilers, and maybe unimplementable as spec'd. I'm working with interested folks in the committee to fix this situation: http://wg21.link/p0190r2

        * wtf/Atomics.h:
        (WTF::zeroWithConsumeDependency): a 0 which carries a dependency
        (WTF::consumeLoad): pixie magic

2016-09-27  JF Bastien  <jfbastien@apple.com>

        Atomics.h on Windows: remove seq_cst hack
        https://bugs.webkit.org/show_bug.cgi?id=162022

        Reviewed by Mark Lam.

        No need to force access to seq_cst, always inlining fixes the MSVC warning.

        * wtf/Atomics.h:
        (WTF::Atomic::compareExchangeWeak): remove seq_cst hack
        (WTF::Atomic::compareExchangeStrong): remove seq_cst hack
        (WTF::Atomic::exchangeAndAdd): remove seq_cst hack
        (WTF::Atomic::exchange): remove seq_cst hack

2016-09-27  Don Olmstead  <don.olmstead@am.sony.com>

        [CMake] Use CMake to determine HAVE_* defines
        https://bugs.webkit.org/show_bug.cgi?id=162368

        Reviewed by Alex Christensen.

        * wtf/Platform.h:

2016-09-20  Anders Carlsson  <andersca@apple.com>

        PlatformEvent::m_modifiers should be an OptionSet
        https://bugs.webkit.org/show_bug.cgi?id=162326

        Reviewed by Daniel Bates.

        * wtf/OptionSet.h:
        (WTF::OptionSet::operator!=):
        (WTF::OptionSet::operator-):

2016-09-27  Jer Noble  <jer.noble@apple.com>

        Remove deprecated ENCRYPTED_MEDIA implementation.
        https://bugs.webkit.org/show_bug.cgi?id=161010

        Reviewed by Eric Carlson.

        Remove ENABLE_ENCRYPTED_MEDIA.

        * wtf/FeatureDefines.h:

2016-09-27  Youenn Fablet  <youenn@apple.com>

        [Fetch API] Use Ref<const T> in FetchBody::m_data variant
        https://bugs.webkit.org/show_bug.cgi?id=162599

        Reviewed by Alex Christensen.

        Enabling to use DeferrableRefCounted<const T> by making m_refCount mutable.

        * wtf/DeferrableRefCounted.h:
        (WTF::DeferrableRefCountedBase::ref):
        (WTF::DeferrableRefCountedBase::derefBase):
        (WTF::DeferrableRefCounted::deref):

2016-09-26  Daniel Bates  <dabates@apple.com>

        Rename IOS_TEXT_AUTOSIZING to TEXT_AUTOSIZING
        https://bugs.webkit.org/show_bug.cgi?id=162365

        Reviewed by Simon Fraser.

        * wtf/FeatureDefines.h:

2016-09-26  Benjamin Poulain  <benjamin@webkit.org>

        [JSC] Shrink the Math inline caches some more
        https://bugs.webkit.org/show_bug.cgi?id=162485

        Reviewed by Saam Barati.

        * wtf/Bag.h:
        Don't copy the arguments before initializing the nodes.

2016-09-26  Michael Catanzaro  <mcatanzaro@igalia.com>

        std::unique_ptr deleter functions should not check if pointer is null
        https://bugs.webkit.org/show_bug.cgi?id=162558

        Reviewed by Alex Christensen.

        std::unique_ptr already does this before calling the deleter.

        * wtf/efl/UniquePtrEfl.h:
        * wtf/glib/GUniquePtr.h:

== Rolled over to ChangeLog-2016-09-26 ==