2017-02-17 JF Bastien <jfbastien@apple.com> A/B test concurrent GC https://bugs.webkit.org/show_bug.cgi?id=168453 <rdar://problem/30553220> Reviewed by Phil Pizlo. Add the SplitTest singleton which can be used for A/B split testing of features. It is currently initialized for WebKit2 WebContent processes, but the class is usable anywhere WTF can be used (it returns an optional, which is valueless if not initialized). * WTF.xcodeproj/project.pbxproj: * wtf/CMakeLists.txt: * wtf/SplitTest.cpp: Added. (WTF::SplitTest::singleton): (WTF::SplitTest::initialize): (WTF::SplitTest::enableBooleanExperiment): * wtf/SplitTest.h: Added. 2017-02-21 Matthew Hanson <matthew_hanson@apple.com> Merge r212692. rdar://problem/30475767 2017-02-20 Mark Lam <mark.lam@apple.com> [Re-landing] CachedCall should let GC know to keep its arguments alive. https://bugs.webkit.org/show_bug.cgi?id=168567 <rdar://problem/30475767> Reviewed by Saam Barati. Added a WTF_FORBID_HEAP_ALLOCATION that will cause a compilation failure if a class declared with it is malloced. While this doesn't prevent that class declared WTF_FORBID_HEAP_ALLOCATION from being embedded in another class that is heap allocated, it does at minimum document the intent and gives the users of this class a chance to do the right thing. * WTF.xcodeproj/project.pbxproj: * wtf/ForbidHeapAllocation.h: Added. 2017-02-20 Matthew Hanson <matthew_hanson@apple.com> Rollout r212660. rdar://problem/30553220 2017-02-20 Matthew Hanson <matthew_hanson@apple.com> Rollout r212646. rdar://problem/30475767 2017-02-17 Matthew Hanson <matthew_hanson@apple.com> A/B test concurrent GC https://bugs.webkit.org/show_bug.cgi?id=168453 <rdar://problem/30553220> Landed on behalf of JF Bastien. Reviewed by Alexey Proskuryakov. Add the SplitTest singleton which can be used for A/B split testing of features. It is currently initialized for WebKit2 WebContent processes, but the class is usable anywhere WTF can be used (it returns an optional, which is valueless if not initialized). * WTF.xcodeproj/project.pbxproj: * wtf/CMakeLists.txt: * wtf/SplitTest.cpp: Added. (WTF::SplitTest::singleton): (WTF::SplitTest::initialize): (WTF::SplitTest::enableBooleanExperiment): * wtf/SplitTest.h: Added. 2017-02-20 Matthew Hanson <matthew_hanson@apple.com> Merge r212618. rdar://problem/30475767 2017-02-19 Mark Lam <mark.lam@apple.com> CachedCall should let GC know to keep its arguments alive. https://bugs.webkit.org/show_bug.cgi?id=168567 <rdar://problem/30475767> Reviewed by Saam Barati. Added a WTF_FORBID_HEAP_ALLOCATION that will cause a compilation failure if a class declared with it is malloced. While this doesn't prevent that class declared WTF_FORBID_HEAP_ALLOCATION from being embedded in another class that is heap allocated, it does at minimum document the intent and gives the users of this class a chance to do the right thing. * WTF.xcodeproj/project.pbxproj: * wtf/ForbidHeapAllocation.h: Added. 2017-02-17 Matthew Hanson <matthew_hanson@apple.com> Merge r212265. rdar://problem/30231732 2017-02-13 Said Abou-Hallawa <sabouhallawa@apple.com> The current frame of an image should not deleted if another frame is asynchronously being decoded https://bugs.webkit.org/show_bug.cgi?id=167618 Reviewed by Simon Fraser. Add ASSERT_IMPLIES() which should fire when a condition is true but the assertion is false. * wtf/Assertions.h: 2017-02-16 Matthew Hanson <matthew_hanson@apple.com> Merge r212256. rdar://problem/30313696 2017-02-13 Brady Eidson <beidson@apple.com> Followup to: Replace all WebKit Library Version checks in WK2 with SDK version checks. https://bugs.webkit.org/show_bug.cgi?id=168124 Reviewed by Geoffrey Garen. * wtf/spi/darwin/dyldSPI.h: 2017-02-16 Matthew Hanson <matthew_hanson@apple.com> Merge r212000. rdar://problem/30313696 2017-02-09 Brady Eidson <beidson@apple.com> Transition "WebKit Library Version" checks to SDK version checks. <rdar://problem/30313696> and https://bugs.webkit.org/show_bug.cgi?id=168056 Reviewed by Geoffrey Garen. * wtf/spi/darwin/dyldSPI.h: Add entries for the iOS 10.3, macOS 10.11, and macOS 10.12.4 SDKs. 2017-02-09 Matthew Hanson <matthew_hanson@apple.com> Merge r211746. rdar://problem/30107622 2017-02-06 Jer Noble <jer.noble@apple.com> Playback stalls when a SourceBuffer append causes frame eviction https://bugs.webkit.org/show_bug.cgi?id=167834 Reviewed by Eric Carlson. Optimize the MediaTime class; specifically the compare() method. The class only needs 6 bits to store the TimeFlags, so make that a uint8_t rather than uint32_t. The implementation is slightly simpler if the TimeScale is unsigned, so make that a uint32_t rather than int32_t. Inline the comparison operators. Optimize the equality comparison by bitwise-and'ing the flags together and masking the result. Optimize for common comparison scenarios (equal timeScales, equal timeValues(), etc.). Attempt the mathematically simpler simpler method for comparing ratios, and only fall back to the complex method if the results of multiplying the timeScale by the timeValue overflows. * wtf/MediaTime.cpp: (WTF::greatestCommonDivisor): (WTF::leastCommonMultiple): (WTF::signum): (WTF::MediaTime::MediaTime): (WTF::MediaTime::createWithFloat): (WTF::MediaTime::createWithDouble): (WTF::MediaTime::operator+): (WTF::MediaTime::operator-): (WTF::MediaTime::operator!): (WTF::MediaTime::operator bool): (WTF::MediaTime::compare): (WTF::MediaTime::setTimeScale): (WTF::abs): (WTF::MediaTime::operator<): Deleted. (WTF::MediaTime::operator>): Deleted. (WTF::MediaTime::operator!=): Deleted. (WTF::MediaTime::operator==): Deleted. (WTF::MediaTime::operator>=): Deleted. (WTF::MediaTime::operator<=): Deleted. * wtf/MediaTime.h: 2017-02-09 Matthew Hanson <matthew_hanson@apple.com> Merge r211622. rdar://problem/30116072 2017-02-02 Andreas Kling <akling@apple.com> [Mac] In-process memory pressure monitor for WebContent processes AKA websam <https://webkit.org/b/167491> <rdar://problem/30116072> Reviewed by Antti Koivisto. Add a WTF helper function for getting the current process's memory footprint. * WTF.xcodeproj/project.pbxproj: * wtf/CMakeLists.txt: * wtf/MemoryFootprint.cpp: (WTF::memoryFootprint): * wtf/MemoryFootprint.h: 2017-02-09 Matthew Hanson <matthew_hanson@apple.com> Merge r211603. rdar://problem/30318237 2017-02-02 Mark Lam <mark.lam@apple.com> Add a SIGILL crash analyzer to make debugging SIGILLs easier. https://bugs.webkit.org/show_bug.cgi?id=167714 <rdar://problem/30318237> Reviewed by Filip Pizlo. * wtf/StdLibExtras.h: 2017-02-05 Matthew Hanson <matthew_hanson@apple.com> Merge r211482. rdar://problem/29711409 2017-02-01 Andreas Kling <akling@apple.com> Implement the alwaysRunsAtBackgroundPriority WK2 setting using thread QoS. <https://webkit.org/b/167387> <rdar://problem/29711409> Reviewed by Antti Koivisto. Add a new mechanism for overriding the max thread QoS level globally: void setGlobalMaxQOSClass(qos_class_t) qos_class_t adjustedQOSClass(qos_class_t) The QoS cap applies to all newly created threads, threads that try to override their QoS class manually, and also passed down to bmalloc. * wtf/Threading.cpp: (WTF::setCurrentThreadIsUserInteractive): (WTF::setCurrentThreadIsUserInitiated): (WTF::setGlobalMaxQOSClass): (WTF::adjustedQOSClass): * wtf/Threading.h: * wtf/ThreadingPthreads.cpp: (WTF::createThreadInternal): * wtf/cocoa/WorkQueueCocoa.cpp: (WTF::dispatchQOSClass): 2017-01-27 Matthew Hanson <matthew_hanson@apple.com> Merge r211237. rdar://problem/30179506 2017-01-26 Saam Barati <sbarati@apple.com> Harden how the compiler references GC objects https://bugs.webkit.org/show_bug.cgi?id=167277 <rdar://problem/30179506> Reviewed by Filip Pizlo. I made TinyPtrSet use bitwise_cast instead of static_cast for its singleEntry() function so that it can work on pointer-like types just as it can on actual pointer types. An example of where this matters is when you have TinyPtrSet<T> where T is defined to be a struct which wraps a pointer, e.g: struct T { void* m_pointer; } * wtf/TinyPtrSet.h: (WTF::TinyPtrSet::singleEntry): 2017-01-26 Matthew Hanson <matthew_hanson@apple.com> Merge r211168. rdar://problem/30154148 2017-01-25 Chris Dumez <cdumez@apple.com> Disable Download attribute support on iOS https://bugs.webkit.org/show_bug.cgi?id=167337 <rdar://problem/30154148> Reviewed by Alexey Proskuryakov. Disable Download attribute support on iOS as it currently does not work. * wtf/FeatureDefines.h: 2017-01-24 Matthew Hanson <matthew_hanson@apple.com> Merge r210603. rdar://problem/29839194 2017-01-11 Anders Carlsson <andersca@apple.com> navigator.plugins.refresh and WKContextRefreshPlugIns doesn't pick up changes to already-present plug-ins https://bugs.webkit.org/show_bug.cgi?id=166942 rdar://problem/29839194 Reviewed by Sam Weinig. * wtf/spi/cf/CFBundleSPI.h: Add SPI declaration. 2017-01-12 Matthew Hanson <matthew_hanson@apple.com> Merge r210399. rdar://problem/29019333 2017-01-05 Myles C. Maxfield <mmaxfield@apple.com> Carets can split up marriages and families https://bugs.webkit.org/show_bug.cgi?id=166711 <rdar://problem/29019333> Reviewed by Alex Christensen. There are four code points which should be allowed to accept emoji modifiers: - U+1F46A FAMILY - U+1F46B MAN AND WOMAN HOLDING HANDS - U+1F46C TWO MEN HOLDING HANDS - U+1F46D TWO WOMEN HOLDING HANDS Even though macOS's and iOS's emoji keyboard don't allow users to actually type these combinations, we may still receive them from other platforms. We should therefore treat these as joining sequences. Rendering isn't a problem because the fonts accept the emoji modifiers, but our caret placement code isn't educated about it. Currently, we treat these emoji groups as ligatures, allowing the caret to be placed between the two code points, which visually shows as being horizontally centered in the glyph. Instead, we should treat these code points as accepting emoji modifiers. Tests: editing/caret/emoji.html editing/caret/ios/emoji.html * wtf/text/TextBreakIterator.cpp: (WTF::cursorMovementIterator): 2017-01-12 Matthew Hanson <matthew_hanson@apple.com> Merge r210609. rdar://problem/27896585 2017-01-11 Andreas Kling <akling@apple.com> Crash when WebCore's GC heap grows way too large. <https://webkit.org/b/166875> <rdar://problem/27896585> Reviewed by Mark Lam. Publish the WTF::GB constant. * wtf/StdLibExtras.h: 2017-01-11 Matthew Hanson <matthew_hanson@apple.com> Merge r210398. rdar://problem/29229439 2017-01-05 Filip Pizlo <fpizlo@apple.com> AutomaticThread timeout shutdown leaves a small window where notify() would think that the thread is still running https://bugs.webkit.org/show_bug.cgi?id=166742 Reviewed by Geoffrey Garen. Remove the use of the RAII ThreadScope, since the use of RAII helped make this bug possible: we'd do ~ThreadScope after we had done ~LockHolder, so in between when we decided to shut down a thread and when it reported itself as being shut down, there was a window where a notify() call would get confused. Now, we run all thread shutdown stuff while the lock is held. We release the lock last. One API implication is that threadWillStop becomes threadIsStopping and it's called while the lock is held. This seems benign. * wtf/AutomaticThread.cpp: (WTF::AutomaticThread::start): (WTF::AutomaticThread::threadIsStopping): (WTF::AutomaticThread::ThreadScope::ThreadScope): Deleted. (WTF::AutomaticThread::ThreadScope::~ThreadScope): Deleted. (WTF::AutomaticThread::threadWillStop): Deleted. * wtf/AutomaticThread.h: 2016-12-19 Babak Shafiei <bshafiei@apple.com> Merge r210010. 2016-12-19 Mark Lam <mark.lam@apple.com> Rolling out r209974 and r209952. They break some websites in mysterious ways. Step 2: Rollout r209952. https://bugs.webkit.org/show_bug.cgi?id=166049 Not reviewed. * wtf/SegmentedVector.h: (WTF::SegmentedVector::last): (WTF::SegmentedVector::first): Deleted. (WTF::SegmentedVector::takeLast): Deleted. 2016-12-19 Babak Shafiei <bshafiei@apple.com> Merge patch for rdar://problem/29466493. 2016-12-19 Dean Jackson <dino@apple.com> Disable some features on the safari-603-branch. <rdar://problem/29466493> - Force the default state of experimental features to off. - Move some experimental features that should be enabled on this branch into the general feature list, so they can't be disabled. - Disable some features that are not ready. * wtf/FeatureDefines.h: 2016-12-16 Mark Lam <mark.lam@apple.com> Add predecessor info to dumps from JSC_dumpBytecodeLivenessResults=true. https://bugs.webkit.org/show_bug.cgi?id=165958 Reviewed by Keith Miller. Added some methods to bring SegmentedVector closer to parity with Vector. * wtf/SegmentedVector.h: (WTF::SegmentedVector::first): (WTF::SegmentedVector::last): (WTF::SegmentedVector::takeLast): 2016-12-16 Michael Saboff <msaboff@apple.com> REGRESSION: HipChat and Mail sometimes hang beneath JSC::Heap::lastChanceToFinalize() https://bugs.webkit.org/show_bug.cgi?id=165962 Reviewed by Filip Pizlo. There is an inherent race in Condition::waitFor() where the timeout can happen just before a notify from another thread. Fixed this by adding a condition variable and flag to each AutomaticThread. The flag is used to signify to a notifying thread that the thread is waiting. That flag is set in the waiting thread before calling waitFor() and cleared by another thread when it notifies the thread. The access to that flag happens when the lock is held. Now the waiting thread checks if the flag after a timeout to see that it in fact should proceed like a normal notification. The added condition variable allows us to target a specific thread. We used to keep a list of waiting threads, now we keep a list of all threads. To notify one thread, we look for a waiting thread and notify it directly. If we can't find a waiting thread, we start a sleeping thread. We notify all threads by waking all waiting threads and starting all sleeping threads. * wtf/AutomaticThread.cpp: (WTF::AutomaticThreadCondition::notifyOne): (WTF::AutomaticThreadCondition::notifyAll): (WTF::AutomaticThread::isWaiting): (WTF::AutomaticThread::notify): (WTF::AutomaticThread::start): * wtf/AutomaticThread.h: 2016-12-15 Myles C. Maxfield <mmaxfield@apple.com> Unreviewed build fix after r209910 Unreviewed. * wtf/Platform.h: 2016-12-15 Myles C. Maxfield <mmaxfield@apple.com> Stop reinterpret_casting UBreakIterators to the undefined type TextBreakIterator https://bugs.webkit.org/show_bug.cgi?id=165931 We have a class declaration for TextBreakIterator but no definition for it. When we create an ICU UBreakIterator, we immediately reinterpret_cast it to this undefined type, and pass it around our code inside WebCore. Then, whenever we want to actually use this iterator, we reinterpret_cast it back to UBreakIterator. This is likely due to some ports historically implementing breaking interators on top of other libraries other than ICU; however, now, all ports use ICU. Because this internal type is not helpful and just adds confusion, we should just call our breaking iterators what they are: UBreakIterators. This patch is a mechanical replacement of TextBreakIterator to UBreakIterator and removes the functions we were calling which pass through directly to ubrk_*(). Reviewed by Alex Christensen. * wtf/text/LineBreakIteratorPoolICU.h: (WTF::LineBreakIteratorPool::take): (WTF::LineBreakIteratorPool::put): * wtf/text/StringView.cpp: (WTF::StringView::GraphemeClusters::Iterator::Impl::computeIndexEnd): * wtf/text/TextBreakIterator.cpp: (WTF::initializeIterator): (WTF::initializeIteratorWithRules): (WTF::setTextForIterator): (WTF::setContextAwareTextForIterator): (WTF::wordBreakIterator): (WTF::sentenceBreakIterator): (WTF::cursorMovementIterator): (WTF::acquireLineBreakIterator): (WTF::releaseLineBreakIterator): (WTF::openLineBreakIterator): (WTF::closeLineBreakIterator): (WTF::getNonSharedCharacterBreakIterator): (WTF::cacheNonSharedCharacterBreakIterator): (WTF::isWordTextBreak): (WTF::numGraphemeClusters): (WTF::numCharactersInGraphemeClusters): (WTF::textBreakFirst): Deleted. (WTF::textBreakLast): Deleted. (WTF::textBreakNext): Deleted. (WTF::textBreakPrevious): Deleted. (WTF::textBreakPreceding): Deleted. (WTF::textBreakFollowing): Deleted. (WTF::textBreakCurrent): Deleted. (WTF::isTextBreak): Deleted. * wtf/text/TextBreakIterator.h: (WTF::LazyLineBreakIterator::lastCharacter): (WTF::LazyLineBreakIterator::secondToLastCharacter): (WTF::LazyLineBreakIterator::setPriorContext): (WTF::LazyLineBreakIterator::updatePriorContext): (WTF::LazyLineBreakIterator::resetPriorContext): (WTF::LazyLineBreakIterator::priorContextLength): (WTF::LazyLineBreakIterator::get): (WTF::NonSharedCharacterBreakIterator::operator UBreakIterator*): (WTF::NonSharedCharacterBreakIterator::operator TextBreakIterator*): Deleted. 2016-12-15 Myles C. Maxfield <mmaxfield@apple.com> Sort Xcode project files https://bugs.webkit.org/show_bug.cgi?id=165937 Reviewed by Simon Fraser. * WTF.xcodeproj/project.pbxproj: 2016-12-15 Konstantin Tokarev <annulen@yandex.ru> Added missing override and final specifiers https://bugs.webkit.org/show_bug.cgi?id=165903 Reviewed by Darin Adler. * wtf/RunLoop.h: 2016-12-15 Yusuke Suzuki <utatane.tea@gmail.com> [JSC] Optimize Kraken stringify https://bugs.webkit.org/show_bug.cgi?id=165857 Reviewed by Darin Adler. Kraken json-stringify-tinderbox performance heavily relies on StringBuilder::appendQuotedJSONString. According to the result produced by Linux `perf`, it occupies 28% of execution time. We tighten the hottest loop in the above function. We create the super fast path for non escaping case. And add " and \ cases (since including " in the string is common). Then we fallback to the slow case. It improves the performance 5.5% in Kraken json-stringify-tinderbox in MBP. Performance result in my MBP (dandelion). Collected 100 samples per benchmark/VM, with 100 VM invocations per benchmark. Emitted a call to gc() between sample measurements. Used 1 benchmark iteration per VM invocation for warm-up. Used the jsc-specific preciseTime() function to get microsecond-level timing. Reporting benchmark execution times with 95% confidence intervals in milliseconds. baseline patched json-stringify-tinderbox 29.243+-0.241 ^ 27.701+-0.235 ^ definitely 1.0557x faster <arithmetic> 29.243+-0.241 ^ 27.701+-0.235 ^ definitely 1.0557x faster Performance result in my Linux laptop (hanayamata). Collected 100 samples per benchmark/VM, with 100 VM invocations per benchmark. Emitted a call to gc() between sample measurements. Used 1 benchmark iteration per VM invocation for warm-up. Used the jsc-specific preciseTime() function to get microsecond-level timing. Reporting benchmark execution times with 95% confidence intervals in milliseconds. baseline patched json-stringify-tinderbox 26.711+-0.475 ^ 25.255+-0.034 ^ definitely 1.0577x faster <arithmetic> 26.711+-0.475 ^ 25.255+-0.034 ^ definitely 1.0577x faster * wtf/text/StringBuilder.cpp: (WTF::appendQuotedJSONStringInternalSlow): (WTF::appendQuotedJSONStringInternal): 2016-12-14 Commit Queue <commit-queue@webkit.org> Unreviewed, rolling out r209795. https://bugs.webkit.org/show_bug.cgi?id=165853 rolled out the wrong revision (Requested by pizlo on #webkit). Reverted changeset: "MarkedBlock::marksConveyLivenessDuringMarking should take into account collection scope" https://bugs.webkit.org/show_bug.cgi?id=165741 http://trac.webkit.org/changeset/209795 2016-12-14 Enrique Ocaña González <eocanha@igalia.com> REGRESSION(r207879-207891): [GStreamer] Introduced many layout test failures and crashes, bots exiting early https://bugs.webkit.org/show_bug.cgi?id=164022 Reviewed by Xabier Rodriguez-Calvar. * wtf/glib/GLibUtilities.h: Added new macros to convert gulong to/from gpointer. 2016-12-14 Gavin Barraclough <barraclough@apple.com> MarkedBlock::marksConveyLivenessDuringMarking should take into account collection scope https://bugs.webkit.org/show_bug.cgi?id=165741 Unreviewed, re-landing this with fix (revert erroneous change to Options). * WTF.xcodeproj/project.pbxproj: * wtf/CMakeLists.txt: * wtf/DataLog.cpp: (WTF::initializeLogFileOnce): (WTF::initializeLogFile): (WTF::dataFile): * wtf/DataLog.h: * wtf/LockedPrintStream.cpp: Added. (WTF::LockedPrintStream::LockedPrintStream): (WTF::LockedPrintStream::~LockedPrintStream): (WTF::LockedPrintStream::vprintf): (WTF::LockedPrintStream::flush): (WTF::LockedPrintStream::begin): (WTF::LockedPrintStream::end): * wtf/LockedPrintStream.h: Added. * wtf/PrintStream.cpp: (WTF::PrintStream::printfVariableFormat): (WTF::PrintStream::begin): (WTF::PrintStream::end): * wtf/PrintStream.h: (WTF::PrintStream::atomically): (WTF::PrintStream::print): (WTF::PrintStream::println): (WTF::PrintStream::printImpl): (WTF::>::unpack): (WTF::FormatImpl::FormatImpl): (WTF::FormatImpl::dump): (WTF::format): (WTF::printInternal): * wtf/RecursiveLockAdapter.h: Added. (WTF::RecursiveLockAdapter::RecursiveLockAdapter): (WTF::RecursiveLockAdapter::lock): (WTF::RecursiveLockAdapter::unlock): (WTF::RecursiveLockAdapter::tryLock): (WTF::RecursiveLockAdapter::isLocked): * wtf/WordLock.cpp: * wtf/WordLock.h: 2016-12-14 Gavin Barraclough <barraclough@apple.com> MarkedBlock::marksConveyLivenessDuringMarking should take into account collection scope https://bugs.webkit.org/show_bug.cgi?id=165741 Unreviewed rollout due to performance regression. * WTF.xcodeproj/project.pbxproj: * wtf/CMakeLists.txt: * wtf/DataLog.cpp: (WTF::initializeLogFileOnce): (WTF::initializeLogFile): (WTF::dataFile): * wtf/DataLog.h: * wtf/LockedPrintStream.cpp: Removed. * wtf/LockedPrintStream.h: Removed. * wtf/PrintStream.cpp: (WTF::PrintStream::printfVariableFormat): Deleted. (WTF::PrintStream::begin): Deleted. (WTF::PrintStream::end): Deleted. * wtf/PrintStream.h: (WTF::PrintStream::print): (WTF::PrintStream::println): (WTF::PrintStream::atomically): Deleted. (WTF::PrintStream::printImpl): Deleted. (): Deleted. (WTF::>::unpack): Deleted. (WTF::FormatImpl::FormatImpl): Deleted. (WTF::FormatImpl::dump): Deleted. (WTF::format): Deleted. * wtf/RecursiveLockAdapter.h: Removed. * wtf/WordLock.cpp: * wtf/WordLock.h: 2016-12-13 JF Bastien <jfbastien@apple.com> std::expected: fix rvalue forwarding issues https://bugs.webkit.org/show_bug.cgi?id=165812 Reviewed by Mark Lam. * wtf/Expected.h: (WTF::UnexpectedType::UnexpectedType): (WTF::ExpectedDetail::Base::Base): (WTF::Expected::Expected): 2016-12-13 Chris Dumez <cdumez@apple.com> Unreviewed, rolling out r209544. Looks like r209489 did not cause the performance regression after all Reverted changeset: "Unreviewed, rolling out r209489." https://bugs.webkit.org/show_bug.cgi?id=165550 http://trac.webkit.org/changeset/209544 2016-12-13 Commit Queue <commit-queue@webkit.org> Unreviewed, rolling out r209725. https://bugs.webkit.org/show_bug.cgi?id=165811 "Broke ARMv7 builds" (Requested by msaboff on #webkit). Reverted changeset: "REGRESSION(r209653): speedometer crashes making virtual slow path tailcalls" https://bugs.webkit.org/show_bug.cgi?id=165748 http://trac.webkit.org/changeset/209725 2016-12-13 JF Bastien <jfbastien@apple.com> [WTF] Turn tryMakeString(), makeString() into variadic templates https://bugs.webkit.org/show_bug.cgi?id=147142 Reviewed by Mark Lam. I wrote this patch while improving WebAssembly's error messages, and only found this bug afterwards. My implementation does the bare minimum to make this code variadic without changing behavior. I think it's better to go with this baby step first, and improve the code later. Notable, for my WebAssembly patch I also taught the code to handle integers and other types (including WebAssembly types). A follow-up could rely on ADL magic to pretty-format these other types. * wtf/text/StringConcatenate.h: (WTF::sumWithOverflow): This unconditionally does the sum for all inputs, which the compiler is more likely to appreciate (because it's the common case) compared to testing for overflow and bailing on each addition (WTF::are8Bit): are: the plural of is! (WTF::makeStringAccumulator): accumulate strings (WTF::tryMakeStringFromAdapters): a small helper which creates the string adapters (WTF::tryMakeString): expose out of WTF, since it's part of this file's used API (WTF::makeString): make it variadic 2016-12-13 Konstantin Tokarev <annulen@yandex.ru> Unreviewed, silence -Wsuggest-attribute for GCC with pragmas * wtf/PrintStream.cpp: (WTF::PrintStream::printfVariableFormat): 2016-12-12 Michael Saboff <msaboff@apple.com> REGRESSION(r209653): speedometer crashes making virtual slow path tailcalls https://bugs.webkit.org/show_bug.cgi?id=165748 Reviewed by Filip Pizlo. Rolling back in r209653, r209654, r209663, and r209673. * wtf/Platform.h: 2016-12-12 Commit Queue <commit-queue@webkit.org> Unreviewed, rolling out r209703. https://bugs.webkit.org/show_bug.cgi?id=165749 Broke Apple builds (Requested by annulen|home on #webkit). Reverted changeset: "Unreviewed, use WTF_ATTRIBUTE_PRINTF instead of clang pragma" http://trac.webkit.org/changeset/209703 2016-12-12 Konstantin Tokarev <annulen@yandex.ru> Unreviewed, use WTF_ATTRIBUTE_PRINTF instead of clang pragma * wtf/PrintStream.cpp: (WTF::PrintStream::printfVariableFormat): * wtf/PrintStream.h: 2016-12-11 Konstantin Tokarev <annulen@yandex.ru> Unreviewed, guarded clang pragma with COMPILER(CLANG) to fix -Werror * wtf/PrintStream.cpp: (WTF::PrintStream::printfVariableFormat): 2016-12-11 Filip Pizlo <fpizlo@apple.com> Change to use #pragma once (requested by Darin Adler). * wtf/RecursiveLockAdapter.h: 2016-12-11 Filip Pizlo <fpizlo@apple.com> Change to use #pragma once (requested by Darin Adler). * wtf/LockedPrintStream.h: 2016-12-10 Filip Pizlo <fpizlo@apple.com> MarkedBlock::marksConveyLivenessDuringMarking should take into account collection scope https://bugs.webkit.org/show_bug.cgi?id=165741 Reviewed by Saam Barati. To find this bug, I needed to seriously beef up our logging infrastructure. It's now the case that: dataLog(...); will print its output atomically. This happens with some careful magic: - dataFile() is now a LockedPrintStream that locks around print(). - The lock is a recursive lock via RecursiveLockAdapter<>, so if the dump methods end up calling back into dataLog() then it just works. This is important: say the dump() calls a getter that itself does logging, maybe because it's encountering badness and wants to report it before crashing). - The lock is a WordLock so that ParkingLot and Lock can keep using dataLog() for debugging. We probably won't need to debug WordLock anytime soon - the algorithm is so simple. - LockedPrintStream::print(...) causes the print callbacks of its arguments to run on the underlying PrintStream, so that you don't need to do recursive lock acquisition on each individual argument and whatever printing it does recursively. * WTF.xcodeproj/project.pbxproj: * wtf/CMakeLists.txt: * wtf/DataLog.cpp: (WTF::initializeLogFileOnce): (WTF::initializeLogFile): (WTF::dataFile): * wtf/DataLog.h: * wtf/LockedPrintStream.cpp: Added. (WTF::LockedPrintStream::LockedPrintStream): (WTF::LockedPrintStream::~LockedPrintStream): (WTF::LockedPrintStream::vprintf): (WTF::LockedPrintStream::flush): (WTF::LockedPrintStream::begin): (WTF::LockedPrintStream::end): * wtf/LockedPrintStream.h: Added. * wtf/PrintStream.cpp: (WTF::PrintStream::printfVariableFormat): (WTF::PrintStream::begin): (WTF::PrintStream::end): * wtf/PrintStream.h: (WTF::PrintStream::atomically): (WTF::PrintStream::print): (WTF::PrintStream::println): (WTF::PrintStream::printImpl): (WTF::>::unpack): (WTF::FormatImpl::FormatImpl): (WTF::FormatImpl::dump): (WTF::format): (WTF::printInternal): * wtf/RecursiveLockAdapter.h: Added. (WTF::RecursiveLockAdapter::RecursiveLockAdapter): (WTF::RecursiveLockAdapter::lock): (WTF::RecursiveLockAdapter::unlock): (WTF::RecursiveLockAdapter::tryLock): (WTF::RecursiveLockAdapter::isLocked): * wtf/WordLock.cpp: * wtf/WordLock.h: 2016-12-11 Darin Adler <darin@apple.com> Use std::vsnprintf instead of vasprintf https://bugs.webkit.org/show_bug.cgi?id=165740 Reviewed by Sam Weinig. * wtf/Platform.h: Remove HAVE_VASPRINTF. * wtf/StringExtras.h: Change the vsnprintf workaround to be used only in older versions of Visual Studio, since the problem it works around was resolved in Visual Studio 2015. 2016-12-10 Commit Queue <commit-queue@webkit.org> Unreviewed, rolling out r209653, r209654, r209663, and r209673. https://bugs.webkit.org/show_bug.cgi?id=165739 speedometer crashes (Requested by pizlo on #webkit). Reverted changesets: "JSVALUE64: Pass arguments in platform argument registers when making JavaScript calls" https://bugs.webkit.org/show_bug.cgi?id=160355 http://trac.webkit.org/changeset/209653 "Unreviewed build fix for 32 bit builds." http://trac.webkit.org/changeset/209654 "Unreviewed build fix for the CLOOP after r209653" http://trac.webkit.org/changeset/209663 "REGRESSION(r209653) Crash in CallFrameShuffler::snapshot()" https://bugs.webkit.org/show_bug.cgi?id=165728 http://trac.webkit.org/changeset/209673 2016-12-10 Chris Dumez <cdumez@apple.com> Avoid calling shrink() in the Vector destructor https://bugs.webkit.org/show_bug.cgi?id=165675 Reviewed by Daniel Bates. Avoid calling shrink() in the Vector destructor to avoid function call overhead and unnecessarily reseting m_size to 0. * wtf/Vector.h: (WTF::Vector::~Vector): (WTF::Vector::asanSetBufferSizeToFullCapacity): (WTF::minCapacity>::asanSetBufferSizeToFullCapacity): 2016-12-09 Michael Saboff <msaboff@apple.com> JSVALUE64: Pass arguments in platform argument registers when making JavaScript calls https://bugs.webkit.org/show_bug.cgi?id=160355 Reviewed by Filip Pizlo. Added a new build option ENABLE_VM_COUNTERS to enable JIT'able counters. The default is for the option to be off. * wtf/Platform.h: Added ENABLE_VM_COUNTERS 2016-12-09 Geoffrey Garen <ggaren@apple.com> Deploy OrdinalNumber in JSC::SourceCode https://bugs.webkit.org/show_bug.cgi?id=165687 Reviewed by Michael Saboff. * wtf/text/OrdinalNumber.h: (WTF::OrdinalNumber::operator>): Added a >. 2016-12-09 Geoffrey Garen <ggaren@apple.com> TextPosition and OrdinalNumber should be more like idiomatic numbers https://bugs.webkit.org/show_bug.cgi?id=165678 Reviewed by Filip Pizlo. * wtf/text/TextPosition.h: (WTF::TextPosition::minimumPosition): Deleted. Just use the default constructor. Other numbers use their default constructors to mean zero. Any time you need a comment that says "used as a default value", that's a pretty good indicator that it should be the default constructor. 2016-12-09 Geoffrey Garen <ggaren@apple.com> Moved OrdinalNumber into its own file https://bugs.webkit.org/show_bug.cgi?id=165663 Reviewed by Saam Barati. * WTF.xcodeproj/project.pbxproj: * wtf/text/OrdinalNumber.h: Copied from Source/WTF/wtf/text/TextPosition.h. (WTF::TextPosition::TextPosition): Deleted. (WTF::TextPosition::operator==): Deleted. (WTF::TextPosition::operator!=): Deleted. (WTF::TextPosition::minimumPosition): Deleted. (WTF::TextPosition::belowRangePosition): Deleted. * wtf/text/TextPosition.h: (WTF::OrdinalNumber::fromZeroBasedInt): Deleted. (WTF::OrdinalNumber::fromOneBasedInt): Deleted. (WTF::OrdinalNumber::OrdinalNumber): Deleted. (WTF::OrdinalNumber::zeroBasedInt): Deleted. (WTF::OrdinalNumber::oneBasedInt): Deleted. (WTF::OrdinalNumber::operator==): Deleted. (WTF::OrdinalNumber::operator!=): Deleted. (WTF::OrdinalNumber::first): Deleted. (WTF::OrdinalNumber::beforeFirst): Deleted. 2016-12-08 David Kilzer <ddkilzer@apple.com> Always check the return value of pthread_key_create() <https://webkit.org/b/165274> Reviewed by Darin Adler. * wtf/ThreadIdentifierDataPthreads.cpp: (WTF::ThreadIdentifierData::initializeOnce): Make the code more readable by assigning a variable to the result of pthread_key_create(). This matches the idiom used elsewhere. 2016-12-08 Keith Miller <keith_miller@apple.com> Add 64-bit signed LEB decode method https://bugs.webkit.org/show_bug.cgi?id=165630 Reviewed by Ryosuke Niwa. Add int64 LEB decode and fix some 64-bit specific issues with the decoder. There is also a fix where we would allow LEBs with canonical length + 1 size that is fixed by this patch. * wtf/LEBDecoder.h: (WTF::LEBDecoder::decodeUInt): (WTF::LEBDecoder::decodeInt): (WTF::LEBDecoder::decodeUInt32): (WTF::LEBDecoder::decodeUInt64): (WTF::LEBDecoder::decodeInt32): (WTF::LEBDecoder::decodeInt64): 2016-12-08 Anders Carlsson <andersca@apple.com> Defer sending Mach messages if the queue is full https://bugs.webkit.org/show_bug.cgi?id=165622 rdar://problem/29518036 Reviewed by Brady Eidson. Add new SPI. * wtf/spi/darwin/XPCSPI.h: 2016-12-06 Filip Pizlo <fpizlo@apple.com> Concurrent GC should be stable enough to land enabled on X86_64 https://bugs.webkit.org/show_bug.cgi?id=164990 Reviewed by Geoffrey Garen. Adds the ability to say: auto locker = holdLock(any type of lock) Instead of having to say: Locker<LockType> locker(locks of type LockType) I think that we should use "auto locker = holdLock(lock)" as the default way that we acquire locks unless we need to use a special locker type. This also adds the ability to safepoint a lock. Safepointing a lock is basically a super fast way of unlocking it fairly and then immediately relocking it - i.e. letting anyone who is waiting to run without losing steam of there is noone waiting. * wtf/Lock.cpp: (WTF::LockBase::safepointSlow): * wtf/Lock.h: (WTF::LockBase::safepoint): * wtf/LockAlgorithm.h: (WTF::LockAlgorithm::safepointFast): (WTF::LockAlgorithm::safepoint): (WTF::LockAlgorithm::safepointSlow): * wtf/Locker.h: (WTF::AbstractLocker::AbstractLocker): (WTF::Locker::tryLock): (WTF::Locker::operator bool): (WTF::Locker::Locker): (WTF::Locker::operator=): (WTF::holdLock): (WTF::tryHoldLock): 2016-12-08 Chris Dumez <cdumez@apple.com> Unreviewed, rolling out r209489. Likely caused large regressions on JetStream, Sunspider and Speedometer Reverted changeset: "Add system trace points for JavaScript VM entry/exit" https://bugs.webkit.org/show_bug.cgi?id=165550 http://trac.webkit.org/changeset/209489 2016-12-07 Simon Fraser <simon.fraser@apple.com> Add system trace points for JavaScript VM entry/exit https://bugs.webkit.org/show_bug.cgi?id=165550 Reviewed by Tim Horton. Add trace points for entry/exit into/out of the JS VM. * wtf/SystemTracing.h: 2016-12-06 Alexey Proskuryakov <ap@apple.com> Correct SDKROOT values in xcconfig files https://bugs.webkit.org/show_bug.cgi?id=165487 rdar://problem/29539209 Reviewed by Dan Bernstein. Fix suggested by Dan Bernstein. * Configurations/DebugRelease.xcconfig: 2016-12-06 Michael Saboff <msaboff@apple.com> REGRESSION(r209399): Causes crashes when dumping JIT disassembly https://bugs.webkit.org/show_bug.cgi?id=165483 Reviewed by Geoffrey Garen. Fixed the RELEASE_ASSERT() to check that the 6 character string is terminated by a null character. * wtf/SixCharacterHash.cpp: (WTF::sixCharacterHashStringToInteger): 2016-12-04 Darin Adler <darin@apple.com> Use ASCIICType more, and improve it a little bit https://bugs.webkit.org/show_bug.cgi?id=165360 Reviewed by Sam Weinig. * wtf/ASCIICType.h: Added declarations of all the functions to the top of the file, so we have a list of what's available, not just a mix of that and the implementation. * wtf/HexNumber.h: (WTF::Internal::hexDigitsForMode): Moved lowerHexDigits and upperHexDigits inside this function. (WTF::appendByteAsHex): Use auto. (WTF::placeByteAsHexCompressIfPossible): Ditto. (WTF::placeByteAsHex): Ditto. (WTF::appendUnsignedAsHex): Ditto. (WTF::appendUnsigned64AsHex): Ditto. (WTF::appendUnsignedAsHexFixedSize): Ditto. (WTF::isHexDigit): Deleted. (WTF::uncheckedHexDigit): Deleted. (WTF::hexDigitValue): Deleted. (WTF::uncheckedHexDigitValue): Deleted. * wtf/SixCharacterHash.cpp: (WTF::sixCharacterHashStringToInteger): Use isASCIIUpper, isASCIILower, and isASCIIDigit. Also added some FIXMEs; for some reason this function uses RELEASE_ASSERT to abort if the passed-in string is not six characters long, and it's not clear to me why this is so critical to assert. (WTF::integerToSixCharacterHashString): Moved the table inside this function, obviating the need for a macro named TABLE. * wtf/dtoa/bignum.cc: (WTF::double_conversion::HexCharValue): Deleted. (WTF::double_conversion::Bignum::AssignHexString): Use toASCIIHexValue. * wtf/dtoa/double-conversion.cc: (WTF::double_conversion::StringToDoubleConverter::StringToDouble): Use isASCIIDigit. * wtf/text/StringBuilder.cpp: (WTF::appendQuotedJSONStringInternal): Use upperNibbleToASCIIHexDigit and lowerNibbleToASCIIHexDigit. * wtf/text/StringImpl.cpp: (WTF::StringImpl::convertToUppercaseWithoutLocale): Use toASCIIUpper. Removed the workaround for a bug that was fixed in Visual Studio 2013. 2016-12-05 Joseph Pecoraro <pecoraro@apple.com> REGRESSION(r208985): SafariForWebKitDevelopment Symbol Not Found looking for method with WTF::Optional https://bugs.webkit.org/show_bug.cgi?id=165351 Reviewed by Yusuke Suzuki. Include a slimmed down version of WTF::Optional which older versions of Safari still depend on for a JavaScriptCore exported symbol. To prevent misuse name it WTF::DeprecatedOptional and use it only in the one instance it is needed. * WTF.xcodeproj/project.pbxproj: * wtf/DeprecatedOptional.h: Added. (WTF::Optional::operator bool): (WTF::Optional::value): (WTF::Optional::asPtr): 2016-12-05 Konstantin Tokarev <annulen@yandex.ru> Add __STDC_FORMAT_MACROS before inttypes.h is included https://bugs.webkit.org/show_bug.cgi?id=165374 We need formatting macros like PRIu64 to be available in all places where inttypes.h header is used. All these usages get inttypes.h definitions via wtf/Assertions.h header, except SQLiteFileSystem.cpp where formatting macros are not used anymore since r185129. This patch fixes multiple build errors with MinGW and reduces number of independent __STDC_FORMAT_MACROS uses in the code base. Reviewed by Darin Adler. * wtf/Assertions.h: Define __STDC_FORMAT_MACROS. * wtf/StdLibExtras.h: Remove definition of PRId64 for Windows, as we have __STDC_FORMAT_MACROS defined now. 2016-12-03 Yusuke Suzuki <utatane.tea@gmail.com> Refactor SymbolImpl layout https://bugs.webkit.org/show_bug.cgi?id=165247 Reviewed by Darin Adler. This patch moves SymbolImpl initialization from StringImpl to SymbolImpl. In SymbolImpl, we create the appropriate fields. At that time, these fields should be aligned to the BufferSubstring StringImpl. And we newly create the `m_flags` in SymbolImpl. Instead of using special StringImpl::null(), we store s_flagIsNullSymbol flag here. In WTF, we have the invariant that StringImpl::empty() is the only atomic empty string. But StringImpl::null() breaks this invariant. Using a special flag is safer way to represent the null Symbol `Symbol()`. * WTF.xcodeproj/project.pbxproj: * wtf/CMakeLists.txt: * wtf/StdLibExtras.h: (WTF::roundUpToMultipleOfImpl0): (WTF::roundUpToMultipleOfImpl): (WTF::roundUpToMultipleOf): * wtf/text/StringImpl.cpp: (WTF::StringImpl::~StringImpl): (WTF::StringImpl::createSymbol): Deleted. (WTF::StringImpl::createNullSymbol): Deleted. * wtf/text/StringImpl.h: (WTF::StringImpl::isAtomic): (WTF::StringImpl::StringImpl): (WTF::StringImpl::requiresCopy): (WTF::StringImpl::isNullSymbol): Deleted. (WTF::StringImpl::symbolAwareHash): Deleted. (WTF::StringImpl::existingSymbolAwareHash): Deleted. (WTF::StringImpl::null): Deleted. (WTF::StringImpl::extractFoldedStringInSymbol): Deleted. (WTF::StringImpl::symbolRegistry): Deleted. (WTF::StringImpl::hashForSymbol): Deleted. * wtf/text/StringStatics.cpp: (WTF::StringImpl::nextHashForSymbol): Deleted. * wtf/text/SymbolImpl.cpp: Copied from Source/WTF/wtf/text/SymbolRegistry.cpp. (WTF::SymbolImpl::nextHashForSymbol): (WTF::SymbolImpl::create): (WTF::SymbolImpl::createNullSymbol): * wtf/text/SymbolImpl.h: (WTF::SymbolImpl::hashForSymbol): (WTF::SymbolImpl::symbolRegistry): (WTF::SymbolImpl::isNullSymbol): (WTF::SymbolImpl::extractFoldedString): (WTF::SymbolImpl::SymbolImpl): (WTF::StringImpl::symbolAwareHash): (WTF::StringImpl::existingSymbolAwareHash): * wtf/text/SymbolRegistry.cpp: (WTF::SymbolRegistry::~SymbolRegistry): (WTF::SymbolRegistry::symbolForKey): (WTF::SymbolRegistry::keyForSymbol): * wtf/text/UniquedStringImpl.h: (WTF::UniquedStringImpl::UniquedStringImpl): 2016-12-01 Yusuke Suzuki <utatane.tea@gmail.com> Introduce StringImpl::StaticStringImpl with constexpr constructor https://bugs.webkit.org/show_bug.cgi?id=165093 Reviewed by Darin Adler. This patch adds new class, StringImpl::StaticStringImpl. By using this class, we can easily create static StringImpls. This class has constexpr constructor. You can initialize instances of this class as global static variables without invoking global constructors. We already have similar system, StaticASCIILiteral. But using it requires some special perl script since we need to calculate hash value. On the other hand, we can use StaticStringImpl without any script supports since we implement constexpr hash function. In the future, we will replace all the use of StaticASCIILiteral with this StaticStringImpl. We define empty / null strings as StaticStringImpl. And we make StringImpl::empty() & StringImpl::null() inline functions. * wtf/Hasher.h: (WTF::StringHasher::hashWithTop8BitsMasked): (WTF::StringHasher::hash): (WTF::StringHasher::finalize): (WTF::StringHasher::finalizeAndMaskTop8Bits): (WTF::StringHasher::computeLiteralHash): (WTF::StringHasher::computeLiteralHashAndMaskTop8Bits): (WTF::StringHasher::avalancheBits3): (WTF::StringHasher::avalancheBits2): (WTF::StringHasher::avalancheBits1): (WTF::StringHasher::avalancheBits0): (WTF::StringHasher::avalancheBits): (WTF::StringHasher::avoidZero): (WTF::StringHasher::processPendingCharacter): (WTF::StringHasher::calculateWithRemainingLastCharacter1): (WTF::StringHasher::calculateWithRemainingLastCharacter0): (WTF::StringHasher::calculateWithRemainingLastCharacter): (WTF::StringHasher::calculate1): (WTF::StringHasher::calculate0): (WTF::StringHasher::calculate): (WTF::StringHasher::computeLiteralHashImpl): * wtf/text/StringImpl.cpp: * wtf/text/StringImpl.h: (WTF::StringImpl::StaticStringImpl::StaticStringImpl): (WTF::StringImpl::null): (WTF::StringImpl::empty): * wtf/text/StringStatics.cpp: (WTF::StringImpl::null): Deleted. (WTF::StringImpl::empty): Deleted. 2016-11-30 Darin Adler <darin@apple.com> Roll out StringBuilder changes from the previous patch. They were a slowdown on a Kraken JSON test. * wtf/text/StringBuilder.cpp: * wtf/text/StringBuilder.h: Roll out changes from below. 2016-11-30 Darin Adler <darin@apple.com> Streamline and speed up tokenizer and segmented string classes https://bugs.webkit.org/show_bug.cgi?id=165003 And do it without re-introducing: REGRESSION (r209058): API test StringBuilderTest.Equal crashing https://bugs.webkit.org/show_bug.cgi?id=165142 Reviewed by Sam Weinig. * wtf/text/StringBuilder.cpp: (WTF::StringBuilder::bufferCharacters<LChar>): Moved this here from the header since it is only used inside the class. Also renamed from getBufferCharacters. (WTF::StringBuilder::bufferCharacters<UChar>): Ditto. (WTF::StringBuilder::appendUninitializedUpconvert): Added. Helper for the upconvert case in the 16-bit overload of StrinBuilder::append. (WTF::StringBuilder::append): Changed to use appendUninitializedUpconvert. (WTF::quotedJSONStringLength): Added. Used in new appendQuotedJSONString implementation below that now correctly determines the size of what will be appended by walking thorugh the string twice. (WTF::appendQuotedJSONStringInternal): Moved the code that writes the quote marks in here. Also made a few coding style tweaks. (WTF::StringBuilder::appendQuotedJSONString): Rewrote to use a much simpler algorithm that grows the string the same way the append function does. The old code would use reserveCapacity in a way that was costly when doing a lot of appends on the same string, and also allocated far too much memory for normal use cases where characters did not need to be turned into escape sequences. * wtf/text/StringBuilder.h: (WTF::StringBuilder::append): Tweaked style a bit, fixed a bug where the m_is8Bit field wasn't set correctly in one case, optimized the function that adds substrings for the case where this is the first append and the substring happens to cover the entire string. Also clarified the assertions and removed an unneeded check from that substring overload. (WTF::equal): Reimplemented, using equalCommon. Added an overload of the equal function just for the case where the arguments are StringBuilder, String. This is needed because of the peculiar behavior of is8Bit in String, different from any of our other string classes. I think we should consider changing String::is8Bit to return true for null strings. We could then remove this overload and probably remove other checks for null and zero length elsewhere that are also needed only to avoid calling is8Bit on a null String. 2016-11-29 JF Bastien <jfbastien@apple.com> WebAssembly JS API: improve Instance https://bugs.webkit.org/show_bug.cgi?id=164757 Reviewed by Keith Miller. * wtf/Expected.h: (WTF::ExpectedDetail::destroy): silence a warning 2016-11-29 Commit Queue <commit-queue@webkit.org> Unreviewed, rolling out r209058 and r209074. https://bugs.webkit.org/show_bug.cgi?id=165188 These changes caused API test StringBuilderTest.Equal to crash and/or fail. (Requested by ryanhaddad on #webkit). Reverted changesets: "Streamline and speed up tokenizer and segmented string classes" https://bugs.webkit.org/show_bug.cgi?id=165003 http://trac.webkit.org/changeset/209058 "REGRESSION (r209058): API test StringBuilderTest.Equal crashing" https://bugs.webkit.org/show_bug.cgi?id=165142 http://trac.webkit.org/changeset/209074 2016-11-29 Simon Fraser <simon.fraser@apple.com> Allow TracePoint to take arbitrary data https://bugs.webkit.org/show_bug.cgi?id=165182 Reviewed by Tim Horton. Allow TracePoint() to take four numeric arguments, for arbitrary data. #pragma once * wtf/SystemTracing.h: (WTF::TracePoint): 2016-11-29 Andy Estes <aestes@apple.com> [Cocoa] Enable two clang warnings recommended by Xcode https://bugs.webkit.org/show_bug.cgi?id=164498 Reviewed by Mark Lam. * Configurations/Base.xcconfig: Enabled CLANG_WARN_INFINITE_RECURSION and CLANG_WARN_SUSPICIOUS_MOVE. 2016-11-29 Darin Adler <darin@apple.com> REGRESSION (r209058): API test StringBuilderTest.Equal crashing https://bugs.webkit.org/show_bug.cgi?id=165142 * wtf/text/StringBuilder.h: Added an overload of the equal function just for the case where the arguments are StringBuilder, String. This is needed because of the peculiar behavior of is8Bit in String, different from any of our other string classes. I think we should consider changing String::is8Bit to return true for null strings. We could then remove this overload and probably remove other checks for null and zero length elsewhere that are also needed only to avoid calling is8Bit on a null String. 2016-11-29 Wenson Hsieh <wenson_hsieh@apple.com> Scroll snapping on Mac should use AppKit animations https://bugs.webkit.org/show_bug.cgi?id=147261 <rdar://problem/29395293> Reviewed by Brent Fulgham. Introduce HAVE(NSSCROLLING_FILTERS), which is on for macOS El Capitan and later. * wtf/Platform.h: 2016-11-28 Darin Adler <darin@apple.com> Streamline and speed up tokenizer and segmented string classes https://bugs.webkit.org/show_bug.cgi?id=165003 Reviewed by Sam Weinig. * wtf/text/StringBuilder.cpp: (WTF::StringBuilder::bufferCharacters<LChar>): Moved this here from the header since it is only used inside the class. Also renamed from getBufferCharacters. (WTF::StringBuilder::bufferCharacters<UChar>): Ditto. (WTF::StringBuilder::appendUninitializedUpconvert): Added. Helper for the upconvert case in the 16-bit overload of StrinBuilder::append. (WTF::StringBuilder::append): Changed to use appendUninitializedUpconvert. (WTF::quotedJSONStringLength): Added. Used in new appendQuotedJSONString implementation below that now correctly determines the size of what will be appended by walking thorugh the string twice. (WTF::appendQuotedJSONStringInternal): Moved the code that writes the quote marks in here. Also made a few coding style tweaks. (WTF::StringBuilder::appendQuotedJSONString): Rewrote to use a much simpler algorithm that grows the string the same way the append function does. The old code would use reserveCapacity in a way that was costly when doing a lot of appends on the same string, and also allocated far too much memory for normal use cases where characters did not need to be turned into escape sequences. * wtf/text/StringBuilder.h: (WTF::StringBuilder::append): Tweaked style a bit, fixed a bug where the m_is8Bit field wasn't set correctly in one case, optimized the function that adds substrings for the case where this is the first append and the substring happens to cover the entire string. Also clarified the assertions and removed an unneeded check from that substring overload. (WTF::equal): Reimplemented, using equalCommon. 2016-11-26 Yusuke Suzuki <utatane.tea@gmail.com> [WTF] Import std::optional reference implementation as WTF::Optional https://bugs.webkit.org/show_bug.cgi?id=164199 Reviewed by Saam Barati and Sam Weinig. Import std::optional reference implementation offered by the C++17 original proposal paper. It has the same interface and functionality to the C++17's one. Previous WTF::Optional lacks several functionalities. The largest one is the correct constexpr constructors. This fact prevents us from using Optional<literal type> in constexpr context in WebKit. In WebKit, we do not allow global constructors. So only the constexpr constructor is the way to use WTF::Optional in the static const global variables. WTF::Optional is used in DOMJIT::Effect and we would like to emit static const global variables that includes this DOMJIT::Effect. That is the main motivation of this work. This functionality allows the IDL code generator to emit DOMJIT signatures as static const global variables. We import the reference implementation one instead of LLVM libc++'s one. This is because LLVM libc++'s one depends on many macro and type_traits offered by libc++ implementation. And adopting it to WebKit requires large modification compared to this reference implementation one. Furthermore, it is difficult to compile libc++'s optional in old GCC and VC++. It also requires some more modifications. To keep the thing simple, we import the reference implementation one now. Once C++17 is released and we update the compiler baseline, we can smoothly switch to the standard library's std::optional. We also add support for the environment that does not use exceptions to this reference implementation. And we also add valueOrCompute helper function. That keeps the extended functionality that previous WTF::Optional has. * wtf/CrossThreadQueue.h: (WTF::CrossThreadQueue<DataType>::tryGetMessage): * wtf/Expected.h: (WTF::makeExpected): * wtf/Forward.h: * wtf/HashTraits.h: (WTF::HashTraits<Ref<P>>::take): * wtf/MainThread.cpp: (WTF::initializeGCThreads): (WTF::mayBeGCThread): * wtf/MainThread.h: * wtf/Optional.h: (std::detail_::is_assignable::has_assign): (std::detail_::has_overloaded_addressof::has_overload): (std::detail_::static_addressof): (std::detail_::convert): (std::nullopt_t::nullopt_t): (std::bad_optional_access::bad_optional_access): (std::optional_base::optional_base): (std::optional_base::~optional_base): (std::constexpr_optional_base::constexpr_optional_base): (std::optional::dataptr): (std::optional::contained_val): (std::optional::__NOEXCEPT_): (std::optional::optional): (std::optional::operator=): (std::optional::emplace): (std::optional::operator ->): (std::optional::operator *): (std::optional::value): (std::optional::value_or): (std::operator==): (std::operator!=): (std::operator<): (std::operator>): (std::operator<=): (std::operator>=): (std::__NOEXCEPT_): (std::make_optional): (std::hash<std::optional<T>>::operator()): (WTF::NulloptTag::NulloptTag): Deleted. (WTF::Optional::Optional): Deleted. (WTF::Optional::~Optional): Deleted. (WTF::Optional::operator=): Deleted. (WTF::Optional::operator bool): Deleted. (WTF::Optional::operator->): Deleted. (WTF::Optional::operator*): Deleted. (WTF::Optional::value): Deleted. (WTF::Optional::valueOr): Deleted. (WTF::Optional::valueOrCompute): Deleted. (WTF::Optional::asPtr): Deleted. (WTF::Optional::destroy): Deleted. (WTF::operator==): Deleted. (WTF::operator!=): Deleted. (WTF::makeOptional): Deleted. (WTF::printInternal): Deleted. * wtf/text/StringView.cpp: (WTF::StringView::GraphemeClusters::Iterator::Impl::Impl): (WTF::StringView::GraphemeClusters::Iterator::Iterator): * wtf/text/StringView.h: 2016-11-26 Simon Fraser <simon.fraser@apple.com> Migrate some layout timer-related code from std::chrono to Seconds and MonotonicTime https://bugs.webkit.org/show_bug.cgi?id=164992 Reviewed by Darin Adler. Add Seconds::zero() as a nicer way to express Seconds(0). * wtf/Seconds.h: (WTF::Seconds::zero): 2016-11-26 Simon Fraser <simon.fraser@apple.com> Add literals for various time units https://bugs.webkit.org/show_bug.cgi?id=165074 Reviewed by Filip Pizlo. Add _min, _s, _ms, _us and _ns literals for easy creation of Seconds from long double and unsigned long long types (those allowed for custom literals). Add minutes-related functions (there are one or two use cases in WebCore). * wtf/Seconds.h: (WTF::Seconds::minutes): (WTF::Seconds::fromMinutes): (WTF::Seconds::fromMilliseconds): (WTF::Seconds::fromMicroseconds): (WTF::Seconds::fromNanoseconds): (WTF::operator _min): (WTF::operator _s): (WTF::operator _ms): (WTF::operator _us): (WTF::operator _ns): 2016-11-22 Darin Adler <darin@apple.com> One more tiny bit of follow-up. * wtf/text/TextBreakIterator.cpp: (WTF::numCharactersInGraphemeClusters): Removed unneeded always-true check. 2016-11-22 Darin Adler <darin@apple.com> Quick follow-up to previous patch. * wtf/text/TextBreakIterator.cpp: (WTF::numCharactersInGraphemeClusters): Removed incorrect assertion. 2016-11-22 Darin Adler <darin@apple.com> Make normal case fast in the input element limitString function https://bugs.webkit.org/show_bug.cgi?id=165023 Reviewed by Dan Bernstein. * wtf/text/LineBreakIteratorPoolICU.h: Removed many unneeded includes. Simplified the class a bit, removing some extra definitions. (WTF::LineBreakIteratorPool::sharedPool): Use NeverDestroyed instead of new. (WTF::LineBreakIteratorPool::makeLocaleWithBreakKeyword): Reimplemented in a simpler way without using StringBuilder. Also updated for change to LineBreakIteratorMode. (WTF::LineBreakIteratorPool::put): Use uncheckedAppend since the code is careful to only use the inline capacity in the vector. * wtf/text/TextBreakIterator.cpp: Moved some includes in here from the header. (WTF::mapLineIteratorModeToRules): Updated for change to LineBreakIteratorMode. (WTF::openLineBreakIterator): Ditto. (WTF::numGraphemeClusters): Added a fast path for all 8-bit strings; don't use ICU for that case, even if there is a CR character in it. (WTF::numCharactersInGraphemeClusters): Added a fast path for strings that are short enough to entirely fit without even looking at the characters; that's a case we likely hit all the time. Also added a fast path for all 8-bit strings. * wtf/text/TextBreakIterator.h: Changed LineBreakIteratorMode to be an enum class and not repeat UAX14 in the names of the modes. Initialize data members in the class definition rather than the constructors. 2016-11-21 Mark Lam <mark.lam@apple.com> Hasher::addCharacters() should be able to handle zero length strings. https://bugs.webkit.org/show_bug.cgi?id=165024 Reviewed by Yusuke Suzuki. Currently, it will fail to handle zero length strings if it has a pending character. The fix is simply to return early if length is 0. * wtf/Hasher.h: (WTF::StringHasher::addCharacters): 2016-11-18 Jeremy Jones <jeremyj@apple.com> Add runtime flag to enable pointer lock. Enable pointer lock feature for mac. https://bugs.webkit.org/show_bug.cgi?id=163801 Reviewed by Simon Fraser. * wtf/FeatureDefines.h: ENABLE_POINTER_LOCK true for Mac. 2016-11-17 Saam Barati <sbarati@apple.com> Remove async/await compile time flag and enable tests https://bugs.webkit.org/show_bug.cgi?id=164828 <rdar://problem/28639334> Reviewed by Yusuke Suzuki. * wtf/FeatureDefines.h: 2016-11-17 Yusuke Suzuki <utatane.tea@gmail.com> [JSC] WTF::TemporaryChange with WTF::SetForScope https://bugs.webkit.org/show_bug.cgi?id=164761 Reviewed by Saam Barati. JavaScriptCore's bytecompiler/SetForScope.h is completely the same to WTF::TemporaryChange. SetForScope sounds better name since it says that this object works as Scope. We rename WTF::TemporaryChange to WTF::SetForScope. And replace all the use to this WTF::SetForScope. * WTF.xcodeproj/project.pbxproj: * wtf/SetForScope.h: Renamed from Source/WTF/wtf/TemporaryChange.h. (WTF::SetForScope::SetForScope): (WTF::SetForScope::~SetForScope): 2016-11-16 Yusuke Suzuki <utatane.tea@gmail.com> [ES6][WebCore] Change ES6_MODULES compile time flag to runtime flag https://bugs.webkit.org/show_bug.cgi?id=164827 Reviewed by Ryosuke Niwa. * wtf/FeatureDefines.h: 2016-11-16 Carlos Alberto Lopez Perez <clopez@igalia.com> [JSC] Build broken for 32-bit x86 after r208306 with GCC 4.9 https://bugs.webkit.org/show_bug.cgi?id=164588 Reviewed by Mark Lam. Provide assembly for executing the cpuid instruction when compiling in PIC mode with the GCC 4.9 EBX on 32-bit x86. Note that the values returned by cpuid here are not used. The purpose of calling this instruction is to force the CPU to complete and sync any buffered modifications on registers, memory or flags before fetching and executing the next instruction. * wtf/Atomics.h: (WTF::x86_cpuid): 2016-11-15 Filip Pizlo <fpizlo@apple.com> Rename CONCURRENT_JIT/ConcurrentJIT to CONCURRENT_JS/ConcurrentJS https://bugs.webkit.org/show_bug.cgi?id=164791 Reviewed by Geoffrey Garen. Both the concurrent GC and the concurrent JIT rely on concurrency support in fundamental JSC runtime components like JSValue. So, the thing that guards it should be a "feature" called CONCURRENT_JS not CONCURRENT_JIT. * wtf/Platform.h: 2016-11-15 Filip Pizlo <fpizlo@apple.com> The concurrent GC should have a timeslicing controller https://bugs.webkit.org/show_bug.cgi?id=164783 Reviewed by Geoffrey Garen. * wtf/LockAlgorithm.h: Added some comments. * wtf/Seconds.h: Added support for modulo. It's necessary for timeslicing. (WTF::Seconds::operator%): (WTF::Seconds::operator%=): 2016-11-11 Filip Pizlo <fpizlo@apple.com> The GC should be optionally concurrent and disabled by default https://bugs.webkit.org/show_bug.cgi?id=164454 Reviewed by Geoffrey Garen. The reason why I went to such great pains to make WTF::Lock fit in two bits is that I knew that I would eventually need to stuff one into some miscellaneous bits of the JSCell header. That time has come, because the concurrent GC has numerous race conditions in visitChildren that can be trivially fixed if each object just has an internal lock. Some cell types might use it to simply protect their entire visitChildren function and anything that mutates the fields it touches, while other cell types might use it as a "lock of last resort" to handle corner cases of an otherwise wait-free or lock-free algorithm. Right now, it's used to protect certain transformations involving indexing storage. To make this happen, I factored the WTF::Lock algorithm into a LockAlgorithm struct that is templatized on lock type (uint8_t for WTF::Lock), the isHeldBit value (1 for WTF::Lock), and the hasParkedBit value (2 for WTF::Lock). This could have been done as a templatized Lock class that basically contains Atomic<LockType>. You could then make any field into a lock by bitwise_casting it to TemplateLock<field type, bit1, bit2>. But this felt too dirty, so instead, LockAlgorithm has static methods that take Atomic<LockType>& as their first argument. I think that this makes it more natural to project a LockAlgorithm onto an existing Atomic<> field. Sadly, some places have to cast their non-Atomic<> field to Atomic<> in order for this to work. Like so many other things we do, this just shows that the C++ style of labeling fields that are subject to atomic ops as atomic is counterproductive. Maybe some day I'll change LockAlgorithm to use our other Atomics API, which does not require Atomic<>. WTF::Lock now uses LockAlgorithm. The slow paths are still outlined. I don't feel too bad about the LockAlgorithm.h header being included in so many places because we change that algorithm so infrequently. Also, I added a hasElapsed(time) function. This function makes it so much more natural to write timeslicing code, which the concurrent GC has to do a lot of. * WTF.xcodeproj/project.pbxproj: * wtf/CMakeLists.txt: * wtf/ListDump.h: * wtf/Lock.cpp: (WTF::LockBase::lockSlow): (WTF::LockBase::unlockSlow): (WTF::LockBase::unlockFairlySlow): (WTF::LockBase::unlockSlowImpl): Deleted. * wtf/Lock.h: (WTF::LockBase::lock): (WTF::LockBase::tryLock): (WTF::LockBase::unlock): (WTF::LockBase::unlockFairly): (WTF::LockBase::isHeld): (): Deleted. * wtf/LockAlgorithm.h: Added. (WTF::LockAlgorithm::lockFastAssumingZero): (WTF::LockAlgorithm::lockFast): (WTF::LockAlgorithm::lock): (WTF::LockAlgorithm::tryLock): (WTF::LockAlgorithm::unlockFastAssumingZero): (WTF::LockAlgorithm::unlockFast): (WTF::LockAlgorithm::unlock): (WTF::LockAlgorithm::unlockFairly): (WTF::LockAlgorithm::isLocked): (WTF::LockAlgorithm::lockSlow): (WTF::LockAlgorithm::unlockSlow): * wtf/TimeWithDynamicClockType.cpp: (WTF::hasElapsed): * wtf/TimeWithDynamicClockType.h: 2016-11-14 JF Bastien <jfbastien@apple.com> Expected: add missing `inline` https://bugs.webkit.org/show_bug.cgi?id=164735 Reviewed by Yusuke Suzuki. Free functions and full template specializations need to be `inline`, or in a .cpp file, otherwise each .o creates a duplicate symbol which makes the linker very sad. * wtf/Expected.h: (WTF::ExpectedDetail::Throw): (WTF::makeExpected): 2016-11-14 Mark Lam <mark.lam@apple.com> Build fix after r208690. https://bugs.webkit.org/show_bug.cgi?id=164681 Not reviewed. * wtf/FastMalloc.h: 2016-11-13 Mark Lam <mark.lam@apple.com> Add debugging facility to limit the max single allocation size. https://bugs.webkit.org/show_bug.cgi?id=164681 Reviewed by Keith Miller. This is useful for simulating memory allocation failures on resource constraint devices for testing purposes. This facility is only conditionally compiled in on debug builds. It does not have any burden on release builds at all. When in use, the max single allocation size limit applies to individual allocations. For malloc (and similar), the allocation will crash in FastMalloc if the requested size exceeds the set max single allocation size. For tryMalloc (and similar), the allocation returns nullptr if the requested size exceeds the set max single allocation size. The max single allocation size is set to std::numeric_limit<size_t>::max() by default (i.e. when not set and no limit is in effect). Also fixed non-bmalloc versions of fastAlignedMalloc() to crash when allocation fails. * wtf/FastMalloc.cpp: (WTF::fastSetMaxSingleAllocationSize): (WTF::fastAlignedMalloc): (WTF::tryFastAlignedMalloc): (WTF::tryFastMalloc): (WTF::fastMalloc): (WTF::tryFastCalloc): (WTF::fastCalloc): (WTF::fastRealloc): * wtf/FastMalloc.h: 2016-11-13 JF Bastien <jfbastien@apple.com> Implement WTF::Expected https://bugs.webkit.org/show_bug.cgi?id=164526 Reviewed by Yusuke Suzuki. std::expected isn't in C++17, and may be in C++20. It's a nice complement to std::any / std::optional because it's a type-tagged union which has a single expected result but could also contain an error. This would be useful in the WebAssembly parser, for example. Using this implementation will allow us to provide feedback to the standards committee and guide std::expected's design before it gets standardized. I've already sent a bunch of feedback to the author based on my experience implementing this. This could supplement WTF::Either and WTF::ExceptionOr. * WTF.xcodeproj/project.pbxproj: * wtf/Compiler.h: Add RELAXED_CONSTEXPR * wtf/Expected.h: Added. (WTF::UnexpectedType::UnexpectedType): (WTF::UnexpectedType::value): (WTF::operator==): (WTF::operator!=): (WTF::operator<): (WTF::operator>): (WTF::operator<=): (WTF::operator>=): (WTF::makeUnexpected): (WTF::ExpectedDetail::Throw): (WTF::ExpectedDetail::ConstexprBase::ConstexprBase): (WTF::ExpectedDetail::Base::Base): (WTF::ExpectedDetail::Base::~Base): (WTF::Expected::Expected): (WTF::Expected::operator=): (WTF::Expected::swap): (WTF::Expected::operator->): (WTF::Expected::operator*): (WTF::Expected::operator bool): (WTF::Expected::hasValue): (WTF::Expected::value): (WTF::Expected::error): (WTF::Expected::getUnexpected): (WTF::Expected::valueOr): (WTF::swap): (WTF::makeExpected): (WTF::makeExpectedFromError): 2016-11-11 Alex Christensen <achristensen@webkit.org> Allow mutable lambdas in HashMap::ensure https://bugs.webkit.org/show_bug.cgi?id=164642 Reviewed by Sam Weinig. * wtf/HashMap.h: (WTF::HashMapEnsureTranslator::translate): (WTF::X>::removeIf): 2016-11-11 Beth Dakin <bdakin@apple.com> Get touch bar code building for open source builds https://bugs.webkit.org/show_bug.cgi?id=164610 Reviewed by Wenson Hsieh. * wtf/Platform.h: 2016-11-10 JF Bastien <jfbastien@apple.com> ASSERTION FAILED: length > offset encountered with wasm.yaml/wasm/js-api/test_Module.js.default-wasm https://bugs.webkit.org/show_bug.cgi?id=164597 Reviewed by Keith Miller. Decoding at end of file should fail, not assert. * wtf/LEBDecoder.h: (WTF::LEBDecoder::decodeUInt): (WTF::LEBDecoder::decodeInt32): 2016-11-10 Alex Christensen <achristensen@webkit.org> Remove unused CFURLCACHE code https://bugs.webkit.org/show_bug.cgi?id=164551 Reviewed by Antti Koivisto. * wtf/Platform.h: 2016-11-09 Alex Christensen <achristensen@webkit.org> Allow RefPtrs of const ThreadSafeRefCounted types https://bugs.webkit.org/show_bug.cgi?id=164548 Reviewed by Tim Horton. * wtf/ThreadSafeRefCounted.h: Make m_refCount mutable like we did with RefCounted in r203257. 2016-11-09 Chris Dumez <cdumez@apple.com> [Mac] Stop using deprecated AppKit enumeration values https://bugs.webkit.org/show_bug.cgi?id=164494 Reviewed by Darin Adler. Stop using deprecated AppKit enumeration values. * wtf/mac/AppKitCompatibilityDeclarations.h: 2016-11-05 Konstantin Tokarev <annulen@yandex.ru> Fixed compilation of LLInt with MinGW https://bugs.webkit.org/show_bug.cgi?id=164449 Reviewed by Michael Catanzaro. MinGW uses LLIntAssembly.h with GNU assembler syntax, just like GCC on other platforms. * wtf/InlineASM.h: Define LOCAL_LABEL_STRING as .L#name for MinGW. 2016-11-05 Konstantin Tokarev <annulen@yandex.ru> [MinGW] Fixed C99/C++11 format attributes in printf-like functions https://bugs.webkit.org/show_bug.cgi?id=164448 Reviewed by Michael Catanzaro. By default MinGW uses printf-like function provided in msvcrt.dll, however they miss support for C99/C++11 format attributes. Use MinGW implementations instead. * wtf/Assertions.h: Use gnu_printf format in WTF_ATTRIBUTE_PRINTF 2016-11-05 Yusuke Suzuki <utatane.tea@gmail.com> [JSCOnly] RunLoopGeneric should adopt MonotonicTime / WallTime change https://bugs.webkit.org/show_bug.cgi?id=164447 Reviewed by Csaba Osztrogonác. Build fix for JSCOnly. * wtf/generic/RunLoopGeneric.cpp: (WTF::RunLoop::TimerBase::ScheduledTask::create): (WTF::RunLoop::TimerBase::ScheduledTask::ScheduledTask): (WTF::RunLoop::TimerBase::ScheduledTask::scheduledTimePoint): (WTF::RunLoop::TimerBase::ScheduledTask::updateReadyTime): (WTF::RunLoop::populateTasks): (WTF::RunLoop::dispatchAfter): (WTF::RunLoop::TimerBase::start): 2016-11-04 Filip Pizlo <fpizlo@apple.com> WTF::ParkingLot should stop using std::chrono because std::chrono::duration casts are prone to overflows https://bugs.webkit.org/show_bug.cgi?id=152045 Reviewed by Andy Estes. We used to use 'double' for all time measurements. Sometimes it was milliseconds, sometimes it was seconds. Sometimes we measured a span of time, sometimes we spoke of time since some epoch. When we spoke of time since epoch, we either used a monotonic clock or a wall clock. The type - always 'double' - never told us what kind of time we had, even though there were roughly six of them (sec interval, ms interval, sec since epoch on wall, ms since epoch on wall, sec since epoch monotonic, ms since epoch monotonic). At some point, we thought that it would be a good idea to replace these doubles with std::chrono. But since replacing some things with std::chrono, we found it to be terribly inconvenient: - Outrageous API. I never want to say std::chrono::milliseconds(blah). I never want to say std::chrono::steady_clock::timepoint. The syntax for duration_cast is ugly, and ideally duration_cast would not even be a thing. - No overflow protection. std::chrono uses integers by default and using anything else is clumsy. But the integer math is done without regard for the rough edges of integer math, so any cast between std::chrono types risks overflow. Any comparison risks overflow because it may do conversions silently. We have even found bugs where some C++ implementations had more overflows than others, which ends up being a special kind of hell. In many cases, the overflow also has nasal demons. It's an error to represent time using integers. It would have been excusable back when floating point math was not guaranteed to be supported on all platforms, but that would have been a long time ago. Time is a continuous, infinite concept and it's a perfect fit for floating point: - Floating point preserves precision under multiplication in all but extreme cases, so using floating point for time means that unit conversions are almost completely lossless. This means that we don't have to think very hard about what units to use. In this patch, we use seconds almost everywhere. We only convert at boundaries, like an API boundary that wants something other than seconds. - Floating point makes it easy to reason about infinity, which is something that time code wants to do a lot. Example: when would you like to timeout? Infinity please! This is the most elegant way of having an API support both a timeout variant and a no-timeout variant. - Floating point does well-understood things when math goes wrong, and these things are pretty well optimized to match what a mathematician would do when computing with real numbers represented using scientific notation with a finite number of significant digits. This means that time math under floating point looks like normal math. On the other hand, std::chrono time math looks like garbage because you have to always check for multiple possible UB corners whenever you touch large integers. Integers that represent time are very likely to be large and you don't have to do much to overflow them. At this time, based on the number of bugs we have already seen due to chrono overflows, I am not certain that we even understand what are all of the corner cases that we should even check for. This patch introduces a new set of timekeeping classes that are all based on double, and all internally use seconds. These classes support algebraic typing. The classes are: - Seconds: this is for measuring a duration. - WallTime: time since epoch according to a wall clock (aka real time clock). - MonotonicTime: time since epoch according to a monotonic clock. - ClockType: enum that says either Wall or Monotonic. - TimeWithDynamicClockType: a tuple of double and ClockType, which represents either a wall time or a monotonic time. All of these classes behave like C++ values and are cheap to copy around since they are very nearly POD. This supports comprehensive conversions between the various time types. Most of this is by way of algebra. Here are just some of the rules we recognize: WallTime = WallTime + Seconds Seconds = WallTime - WallTime MonotonicTime = MonotonicTime + Seconds etc... We support negative, infinite, and NaN times because math. We support conversions between MonotonicTime and WallTime, like: WallTime wt = mt.approximateWallTime() This is called this "approximate" because the only way to do it is to get the current time on both clocks and convert relative to that. Many of our APIs would be happy using whatever notion of time the user wanted to use. For those APIs, which includes Condition and ParkingLot, we have TimeWithDynamicClockType. You can automatically convert WallTime or MonotonicTime to TimeWithDynamicClockType. This means that if you use a WallTime with Condition::waitUntil, then Condition's internal logic for when it should wake up makes its decision based on the current WallTime - but if you use MonotonicTime then waitUntil will make its decision based on current MonotonicTime. This is a greater level of flexibility than chrono allowed, since chrono did not have the concept of a dynamic clock type. This patch does not include conversions between std::chrono and these new time classes, because past experience shows that we're quite bad at getting conversions between std::chrono and anything else right. Also, I didn't need such conversion code because this patch only converts code that transitively touches ParkingLot and Condition. It was easy to get all of that code onto the new time classes. * WTF.xcodeproj/project.pbxproj: * wtf/AutomaticThread.cpp: (WTF::AutomaticThread::start): * wtf/CMakeLists.txt: * wtf/ClockType.cpp: Added. (WTF::printInternal): * wtf/ClockType.h: Added. * wtf/Condition.h: (WTF::ConditionBase::waitUntil): (WTF::ConditionBase::waitFor): (WTF::ConditionBase::wait): (WTF::ConditionBase::waitUntilWallClockSeconds): Deleted. (WTF::ConditionBase::waitUntilMonotonicClockSeconds): Deleted. (WTF::ConditionBase::waitForSeconds): Deleted. (WTF::ConditionBase::waitForSecondsImpl): Deleted. (WTF::ConditionBase::waitForImpl): Deleted. (WTF::ConditionBase::absoluteFromRelative): Deleted. * wtf/CrossThreadQueue.h: (WTF::CrossThreadQueue<DataType>::waitForMessage): * wtf/CurrentTime.cpp: (WTF::sleep): * wtf/MessageQueue.h: (WTF::MessageQueue::infiniteTime): Deleted. * wtf/MonotonicTime.cpp: Added. (WTF::MonotonicTime::now): (WTF::MonotonicTime::approximateWallTime): (WTF::MonotonicTime::dump): (WTF::MonotonicTime::sleep): * wtf/MonotonicTime.h: Added. (WTF::MonotonicTime::MonotonicTime): (WTF::MonotonicTime::fromRawDouble): (WTF::MonotonicTime::infinity): (WTF::MonotonicTime::secondsSinceEpoch): (WTF::MonotonicTime::approximateMonotonicTime): (WTF::MonotonicTime::operator bool): (WTF::MonotonicTime::operator+): (WTF::MonotonicTime::operator-): (WTF::MonotonicTime::operator+=): (WTF::MonotonicTime::operator-=): (WTF::MonotonicTime::operator==): (WTF::MonotonicTime::operator!=): (WTF::MonotonicTime::operator<): (WTF::MonotonicTime::operator>): (WTF::MonotonicTime::operator<=): (WTF::MonotonicTime::operator>=): * wtf/ParkingLot.cpp: (WTF::ParkingLot::parkConditionallyImpl): (WTF::ParkingLot::unparkOne): (WTF::ParkingLot::unparkOneImpl): (WTF::ParkingLot::unparkCount): * wtf/ParkingLot.h: (WTF::ParkingLot::parkConditionally): (WTF::ParkingLot::compareAndPark): * wtf/Seconds.cpp: Added. (WTF::Seconds::operator+): (WTF::Seconds::operator-): (WTF::Seconds::dump): (WTF::Seconds::sleep): * wtf/Seconds.h: Added. (WTF::Seconds::Seconds): (WTF::Seconds::value): (WTF::Seconds::seconds): (WTF::Seconds::milliseconds): (WTF::Seconds::microseconds): (WTF::Seconds::nanoseconds): (WTF::Seconds::fromMilliseconds): (WTF::Seconds::fromMicroseconds): (WTF::Seconds::fromNanoseconds): (WTF::Seconds::infinity): (WTF::Seconds::operator bool): (WTF::Seconds::operator+): (WTF::Seconds::operator-): (WTF::Seconds::operator*): (WTF::Seconds::operator/): (WTF::Seconds::operator+=): (WTF::Seconds::operator-=): (WTF::Seconds::operator*=): (WTF::Seconds::operator/=): (WTF::Seconds::operator==): (WTF::Seconds::operator!=): (WTF::Seconds::operator<): (WTF::Seconds::operator>): (WTF::Seconds::operator<=): (WTF::Seconds::operator>=): * wtf/TimeWithDynamicClockType.cpp: Added. (WTF::TimeWithDynamicClockType::now): (WTF::TimeWithDynamicClockType::nowWithSameClock): (WTF::TimeWithDynamicClockType::wallTime): (WTF::TimeWithDynamicClockType::monotonicTime): (WTF::TimeWithDynamicClockType::approximateWallTime): (WTF::TimeWithDynamicClockType::approximateMonotonicTime): (WTF::TimeWithDynamicClockType::operator-): (WTF::TimeWithDynamicClockType::operator<): (WTF::TimeWithDynamicClockType::operator>): (WTF::TimeWithDynamicClockType::operator<=): (WTF::TimeWithDynamicClockType::operator>=): (WTF::TimeWithDynamicClockType::dump): (WTF::TimeWithDynamicClockType::sleep): * wtf/TimeWithDynamicClockType.h: Added. (WTF::TimeWithDynamicClockType::TimeWithDynamicClockType): (WTF::TimeWithDynamicClockType::fromRawDouble): (WTF::TimeWithDynamicClockType::secondsSinceEpoch): (WTF::TimeWithDynamicClockType::clockType): (WTF::TimeWithDynamicClockType::withSameClockAndRawDouble): (WTF::TimeWithDynamicClockType::operator bool): (WTF::TimeWithDynamicClockType::operator+): (WTF::TimeWithDynamicClockType::operator-): (WTF::TimeWithDynamicClockType::operator+=): (WTF::TimeWithDynamicClockType::operator-=): (WTF::TimeWithDynamicClockType::operator==): (WTF::TimeWithDynamicClockType::operator!=): * wtf/WallTime.cpp: Added. (WTF::WallTime::now): (WTF::WallTime::approximateMonotonicTime): (WTF::WallTime::dump): (WTF::WallTime::sleep): * wtf/WallTime.h: Added. (WTF::WallTime::WallTime): (WTF::WallTime::fromRawDouble): (WTF::WallTime::infinity): (WTF::WallTime::secondsSinceEpoch): (WTF::WallTime::approximateWallTime): (WTF::WallTime::operator bool): (WTF::WallTime::operator+): (WTF::WallTime::operator-): (WTF::WallTime::operator+=): (WTF::WallTime::operator-=): (WTF::WallTime::operator==): (WTF::WallTime::operator!=): (WTF::WallTime::operator<): (WTF::WallTime::operator>): (WTF::WallTime::operator<=): (WTF::WallTime::operator>=): * wtf/threads/BinarySemaphore.cpp: (WTF::BinarySemaphore::wait): * wtf/threads/BinarySemaphore.h: 2016-11-03 Filip Pizlo <fpizlo@apple.com> DFG plays fast and loose with the shadow values of a Phi https://bugs.webkit.org/show_bug.cgi?id=164309 Reviewed by Saam Barati. Made this API use size rather than maxIndex as its initialization parameter, because that's less confusing. * wtf/IndexSparseSet.h: (WTF::IndexSparseSet<OverflowHandler>::IndexSparseSet): 2016-11-03 Commit Queue <commit-queue@webkit.org> Unreviewed, rolling out r208364. https://bugs.webkit.org/show_bug.cgi?id=164402 broke the build (Requested by smfr on #webkit). Reverted changeset: "DFG plays fast and loose with the shadow values of a Phi" https://bugs.webkit.org/show_bug.cgi?id=164309 http://trac.webkit.org/changeset/208364 2016-11-03 Filip Pizlo <fpizlo@apple.com> DFG plays fast and loose with the shadow values of a Phi https://bugs.webkit.org/show_bug.cgi?id=164309 Reviewed by Saam Barati. Made this API use size rather than maxIndex as its initialization parameter, because that's less confusing. * wtf/IndexSparseSet.h: (WTF::IndexSparseSet<OverflowHandler>::IndexSparseSet): 2016-11-03 Konstantin Tokarev <annulen@yandex.ru> Fixes to build JSCOnly on macOS https://bugs.webkit.org/show_bug.cgi?id=164379 Reviewed by Michael Catanzaro. * wtf/Platform.h: JSCOnly port should not provide any PLATFORM() macro 2016-11-03 Brady Eidson <beidson@apple.com> IndexedDB 2.0: Support binary keys. <rdar://problem/28806927> and https://bugs.webkit.org/show_bug.cgi?id=164359 Reviewed by Alex Christensen. * wtf/Hasher.h: (WTF::StringHasher::hashMemory): Teach hashMemory() to handle buffers with odd lengths. 2016-11-02 Filip Pizlo <fpizlo@apple.com> The GC should be in a thread https://bugs.webkit.org/show_bug.cgi?id=163562 Reviewed by Geoffrey Garen and Andreas Kling. This fixes some bugs and adds a few features. * wtf/Atomics.h: The GC may do work on behalf of the JIT. If it does, the main thread needs to execute a cross-modifying code fence. This is cpuid on x86 and I believe it's isb on ARM. It would have been an isync on PPC and I think that isb is the ARM equivalent. (WTF::arm_isb): (WTF::crossModifyingCodeFence): (WTF::x86_ortop): (WTF::x86_cpuid): * wtf/AutomaticThread.cpp: I accidentally had AutomaticThreadCondition inherit from ThreadSafeRefCounted<AutomaticThread> [sic]. This never crashed before because all of our prior AutomaticThreadConditions were immortal. (WTF::AutomaticThread::AutomaticThread): (WTF::AutomaticThread::~AutomaticThread): (WTF::AutomaticThread::start): * wtf/AutomaticThread.h: * wtf/MainThread.cpp: Need to allow initializeGCThreads() to be called separately because it's now more than just a debugging thing. (WTF::initializeGCThreads): 2016-11-02 Carlos Alberto Lopez Perez <clopez@igalia.com> Clean wrong comment about compositing on the UI process. https://bugs.webkit.org/show_bug.cgi?id=164339 Reviewed by Michael Catanzaro. * wtf/Platform.h: The comment about compositing on the UI process was added on r109302 but was not removed properly when the Qt port was removed from trunk. USE_PROTECTION_SPACE_AUTH_CALLBACK has nothing to do with it. 2016-11-02 Alex Christensen <achristensen@webkit.org> Remove Battery Status API from the tree https://bugs.webkit.org/show_bug.cgi?id=164213 Reviewed by Sam Weinig. * wtf/FeatureDefines.h: 2016-11-02 Romain Bellessort <romain.bellessort@crf.canon.fr> [Readable Streams API] Enable creation of ReadableByteStreamController https://bugs.webkit.org/show_bug.cgi?id=164014 Reviewed by Youenn Fablet. Added flag for the byte stream part of Readable Streams API. * wtf/FeatureDefines.h: 2016-11-02 Per Arne Vollan <pvollan@apple.com> [Win] Copy build results to AAS 'Program Files' folder. https://bugs.webkit.org/show_bug.cgi?id=164273 Reviewed by Brent Fulgham. The preferred location for the binaries is the AAS 'Program Files' folder. * WTF.vcxproj/WTF.proj: 2016-10-29 Filip Pizlo <fpizlo@apple.com> JSC should support SharedArrayBuffer https://bugs.webkit.org/show_bug.cgi?id=163986 Reviewed by Keith Miller. Adds some small things we need for SharedArrayBuffer. * wtf/Atomics.h: (WTF::Atomic::compareExchangeWeakRelaxed): (WTF::Atomic::exchangeAdd): (WTF::Atomic::exchangeAnd): (WTF::Atomic::exchangeOr): (WTF::Atomic::exchangeSub): (WTF::Atomic::exchangeXor): (WTF::atomicLoad): (WTF::atomicStore): (WTF::atomicCompareExchangeWeak): (WTF::atomicCompareExchangeWeakRelaxed): (WTF::atomicCompareExchangeStrong): (WTF::atomicExchangeAdd): (WTF::atomicExchangeAnd): (WTF::atomicExchangeOr): (WTF::atomicExchangeSub): (WTF::atomicExchangeXor): (WTF::atomicExchange): (WTF::Atomic::exchangeAndAdd): Deleted. (WTF::weakCompareAndSwap): Deleted. We need to be able to do atomics operations on naked pointers. We also need to be able to do all of the things that std::atomic does. This adds those things and renames weakCompareAndSwap to atomicCompareExchangeWeakRelaxed so that we're using consistent terminology. * wtf/Bitmap.h: (WTF::WordType>::concurrentTestAndSet): Renamed weakCompareAndSwap. (WTF::WordType>::concurrentTestAndClear): Renamed weakCompareAndSwap. * wtf/FastBitVector.h: (WTF::FastBitVector::atomicSetAndCheck): Renamed weakCompareAndSwap. * wtf/ParkingLot.cpp: (WTF::ParkingLot::unparkOne): (WTF::ParkingLot::unparkCount): * wtf/ParkingLot.h: Added unparkCount(), which lets you unpark some bounded number of threads and returns the number of threads unparked. This is just a modest extension of unparkAll(). unparkAll() now just calls unparkCount(ptr, UINT_MAX). 2016-10-30 Frederic Wang <fwang@igalia.com> Use HarfBuzz ot-math API to parse the OpenType MATH table https://bugs.webkit.org/show_bug.cgi?id=162671 Reviewed by Michael Catanzaro. * wtf/Platform.h: By default, do not enable internal OpenType MATH parsing on GTK. 2016-10-25 Mark Lam <mark.lam@apple.com> String.prototype.replace() should throw an OutOfMemoryError when using too much memory. https://bugs.webkit.org/show_bug.cgi?id=163996 <rdar://problem/28263117> Reviewed by Geoffrey Garen. * wtf/Vector.h: (WTF::minCapacity>::tryConstructAndAppend): (WTF::minCapacity>::tryConstructAndAppendSlowCase): - Added try versions of constructAndAppend() so that we can handle the failure to allocate more gracefully. 2016-10-25 Konstantin Tokarev <annulen@yandex.ru> Non-specialized version of deleteObject should not have template argument https://bugs.webkit.org/show_bug.cgi?id=163943 Reviewed by Anders Carlsson. Fixes compilation of GDIObject.h with MinGW * wtf/win/GDIObject.h: (WTF::deleteObject): 2016-10-24 Per Arne Vollan <pvollan@apple.com> [Win] CMake build type is not set. https://bugs.webkit.org/show_bug.cgi?id=163917 Reviewed by Alex Christensen. The CMAKE_BUILD_TYPE variable should be set to Debug or Release. * WTF.vcxproj/WTF.proj: 2016-10-23 Yusuke Suzuki <utatane.tea@gmail.com> [DOMJIT] Add a way for DOMJIT::Patchpoint to express effects https://bugs.webkit.org/show_bug.cgi?id=163657 Reviewed by Saam Barati. Simplify nonEmptyRangesOverlap. * wtf/MathExtras.h: (WTF::nonEmptyRangesOverlap): 2016-10-23 Chris Dumez <cdumez@apple.com> Another unreviewed attempt to fix the WatchOS / TvOS build after r207585. <rdar://problem/28902292> Disable USE_CFURLCONNECTION on newer WatchOS / TvOS. * wtf/Platform.h: 2016-10-20 Keith Miller <keith_miller@apple.com> Add support for WASM calls https://bugs.webkit.org/show_bug.cgi?id=161727 Reviewed by Filip Pizlo and Michael Saboff. Added a new decodeUInt64. Also, added WTF::LEBDecoder namespace. * wtf/LEBDecoder.h: (WTF::LEBDecoder::decodeUInt): (WTF::LEBDecoder::decodeUInt32): (WTF::LEBDecoder::decodeUInt64): (WTF::LEBDecoder::decodeInt32): (decodeUInt32): Deleted. (decodeInt32): Deleted. 2016-10-20 Filip Pizlo <fpizlo@apple.com> The tracking of the coarse-grain Heap state (allocating or not, collector or not, eden vs full) should respect the orthogonality between allocating and collecting https://bugs.webkit.org/show_bug.cgi?id=163738 Reviewed by Geoffrey Garen. There will soon be different kinds of GC threads, and WTF's "are you a GC thread" thing should know about this. * wtf/MainThread.cpp: (WTF::initializeGCThreads): (WTF::registerGCThread): (WTF::mayBeGCThread): * wtf/MainThread.h: 2016-10-19 Sam Weinig <sam@webkit.org> Add convenience function that combines WTF::visit(...) with WTF::makeVisitor(...) https://bugs.webkit.org/show_bug.cgi?id=163713 Reviewed by Dan Bernstein. - Add WTF::switchOn which merges WTF::visit with WTF::makeVisitor in the following way: WTF::visit(WTF::makeVisitor(...), variant) * wtf/Variant.h: (WTF::switchOn): 2016-10-19 Alex Christensen <achristensen@webkit.org> Revert r207151 https://bugs.webkit.org/show_bug.cgi?id=163675 Reviewed by Brent Fulgham. * wtf/Platform.h: * wtf/SchedulePair.h: * wtf/SchedulePairMac.mm: 2016-10-19 Filip Pizlo <fpizlo@apple.com> REGRESSION (r207480): 3 Dromaeo tests failing https://bugs.webkit.org/show_bug.cgi?id=163633 Reviewed by Mark Lam. It's a ParkingLot bug: if we timeout and get unparked at the same time, then the unparking thread will clear our address eventually - but not immediately. This causes a nasty assertion failure. The tricky thing is that when we detect this, we need to wait until that unparking thread does the deed. Otherwise, they will still do it at some later time. Alternatively, we could use some kind of versioning to detect this - increment the version when you park, so that unparking threads will know if they are time travelers. That seems more yucky. I don't think it matters too much what we do here, so long as it's simple and correct, since this requires a race that is rare enough that it didn't happen once in my testing, and I was pretty thorough. For example, it didn't happen once in 15 runs of JetStream. The race is rare because it requires the timeout to happen right as someone else is unparking. Since it's so rare, its probably OK that the unparked thread waits just a tiny moment until the unparking thread is done. * wtf/ParkingLot.cpp: (WTF::ParkingLot::parkConditionallyImpl): (WTF::ParkingLot::unparkOneImpl): 2016-10-19 Filip Pizlo <fpizlo@apple.com> Baseline JIT should use AutomaticThread https://bugs.webkit.org/show_bug.cgi?id=163686 Reviewed by Geoffrey Garen. Added a AutomaticThreadCondition::wait() method, so that if you really want to use one common condition for your thread and something else, you can do it. This trivially works if you only use notifyAll(), and behaves as you'd expect for notifyOne() (i.e. it's dangerous, since you don't know who will wake up). The Baseline JIT used the one-true-Condition idiom because it used notifyAll() in an optimal way: there are just two threads talking to each other, so it wakes up at most one thread and that thread is exactly the one you want woken up. Adding wait() means that I did not have to change that code. * wtf/AutomaticThread.cpp: (WTF::AutomaticThreadCondition::wait): * wtf/AutomaticThread.h: 2016-10-18 Filip Pizlo <fpizlo@apple.com> DFG worklist should use AutomaticThread https://bugs.webkit.org/show_bug.cgi?id=163615 Reviewed by Mark Lam. This adds new functionality to AutomaticThread to support DFG::Worklist: - AutomaticThread::threadDidStart/threadWillStop virtual methods called at the start and end of a thread's lifetime. This allows Worklist to tie some resources to the life of the thread, and also means that now those resources will naturally free up when the Worklist is not in use. - AutomaticThreadCondition::notifyOne(). This required changes to Condition::notifyOne(). We need to know if the Condition woke up anyone. If it didn't, then we need to launch one of our threads. * wtf/AutomaticThread.cpp: (WTF::AutomaticThreadCondition::notifyOne): (WTF::AutomaticThread::ThreadScope::ThreadScope): (WTF::AutomaticThread::ThreadScope::~ThreadScope): (WTF::AutomaticThread::start): (WTF::AutomaticThread::threadDidStart): (WTF::AutomaticThread::threadWillStop): * wtf/AutomaticThread.h: * wtf/Condition.h: (WTF::ConditionBase::notifyOne): 2016-10-18 Sam Weinig <sam@webkit.org> Replace std::experimental::variant with WTF::Variant (or similar) https://bugs.webkit.org/show_bug.cgi?id=163626 Reviewed by Chris Dumez. Rename std::experimental::variant, Variant. Move helpers get/holds_alternative/etc. into the WTF namespace. * wtf/Forward.h: * wtf/Variant.h: 2016-10-18 Filip Pizlo <fpizlo@apple.com> WTF should make it easier to create threads that die automatically after inactivity https://bugs.webkit.org/show_bug.cgi?id=163576 Reviewed by Andreas Kling. For a long time now, I've been adding threads to WTF/JSC and each time I do this, I feel guilty because those threads don't shut down when they are inactive. For example, in bug 163562, I need to add a new GC thread. There will be one of them per VM. This means that a JSC API client that starts a lot of VMs will have a lot of threads. I don't think that's good. A common pattern for all of these threads is that they have some well-defined trigger that causes them to run. This trigger has a lock, a condition variable, some logic that determines if there is work to do, and then of course the logic for the thread's actual work. The thread bodies usually look like this: void Thingy::runThread() { for (;;) { Work work; { LockHolder locker(m_lock); while (!hasWork()) m_cond.wait(m_lock); work = takeWork(); } doWork(work); } } If you look at ParallelHelperPool (the GC's threads) and DFG::Worklist (some of the JIT's threads), you will see this pattern. This change adds a new kind of thread, called AutomaticThread, that lets you write threads to this pattern while getting automatic thread shutdown for free: instead of just waiting on a condition variable, AutomaticThread will have a timeout that causes the thread to die. The condition variable associated with AutomaticThread, called AutomaticThreadCondition, is smart enough to restart any threads that have decided to stop due to inactivity. The inactivity threshold is current just 1 second. In this patch I only adopt AutomaticThread for ParallelHelperPool. I plan to adopt it in more places soon. * WTF.xcodeproj/project.pbxproj: * wtf/AutomaticThread.cpp: Added. (WTF::AutomaticThreadCondition::create): (WTF::AutomaticThreadCondition::AutomaticThreadCondition): (WTF::AutomaticThreadCondition::~AutomaticThreadCondition): (WTF::AutomaticThreadCondition::notifyAll): (WTF::AutomaticThreadCondition::add): (WTF::AutomaticThreadCondition::remove): (WTF::AutomaticThreadCondition::contains): (WTF::AutomaticThread::AutomaticThread): (WTF::AutomaticThread::~AutomaticThread): (WTF::AutomaticThread::join): (WTF::AutomaticThread::start): * wtf/AutomaticThread.h: Added. * wtf/CMakeLists.txt: * wtf/ParallelHelperPool.cpp: (WTF::ParallelHelperClient::ParallelHelperClient): (WTF::ParallelHelperClient::~ParallelHelperClient): (WTF::ParallelHelperClient::setTask): (WTF::ParallelHelperClient::finish): (WTF::ParallelHelperClient::doSomeHelping): (WTF::ParallelHelperClient::runTask): (WTF::ParallelHelperPool::ParallelHelperPool): (WTF::ParallelHelperPool::~ParallelHelperPool): (WTF::ParallelHelperPool::ensureThreads): (WTF::ParallelHelperPool::doSomeHelping): (WTF::ParallelHelperPool::Thread::Thread): (WTF::ParallelHelperPool::didMakeWorkAvailable): (WTF::ParallelHelperPool::helperThreadBody): Deleted. (WTF::ParallelHelperPool::waitForClientWithTask): Deleted. * wtf/ParallelHelperPool.h: 2016-10-18 Said Abou-Hallawa <sabouhallawa@apple.com> SVGCSSParser: m_implicitShorthand value is not reset after adding the shorthand property https://bugs.webkit.org/show_bug.cgi?id=116470 Reviewed by Simon Fraser. * wtf/TemporaryChange.h: (WTF::TemporaryChange::TemporaryChange): Add a new constructor to make TemporaryChange work as a restorer. The temporary change will happen after we construct the object. 2016-10-17 Simon Fraser <simon.fraser@apple.com> Implement DOMRect/DOMRectReadOnly https://bugs.webkit.org/show_bug.cgi?id=163464 Reviewed by Darin Adler. Implement min()/max() in a way that follows Math.min/Math.max, which return NaN if either argument is NaN. * wtf/MathExtras.h: (WTF::nanPropagatingMin): (WTF::nanPropagatingMax): 2016-10-15 Sam Weinig <sam@webkit.org> MessageEvent's source property should be a (DOMWindow or MessagePort)? rather than a EventTarget? https://bugs.webkit.org/show_bug.cgi?id=163475 Reviewed by Simon Fraser. * wtf/Variant.h: Add missing return statement that was tripping up some compilers. 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 ==