diff options
Diffstat (limited to 'JavaScriptCore/ChangeLog')
| -rw-r--r-- | JavaScriptCore/ChangeLog | 1202 |
1 files changed, 1202 insertions, 0 deletions
diff --git a/JavaScriptCore/ChangeLog b/JavaScriptCore/ChangeLog index eb2f97d..85755cc 100644 --- a/JavaScriptCore/ChangeLog +++ b/JavaScriptCore/ChangeLog @@ -1,3 +1,1205 @@ +2010-05-17 Patrick Gansterer <paroga@paroga.com> + + Reviewed by Laszlo Gombos. + + [Qt] Remove WinCE endian dedection. + https://bugs.webkit.org/show_bug.cgi?id=38511 + + Windows CE supports little-endian format only. + Correct dedection was added in r57804. + + * wtf/Platform.h: + +2010-05-16 Simon Fraser <simon.fraser@apple.com> + + Reviewed by Darin Adler. + + Reduce the size of FunctionExecutable + https://bugs.webkit.org/show_bug.cgi?id=39180 + + Change m_numVariables from a size_t to 31 bits in a bitfield, + packed with another bit for m_forceUsesArguments (which in turn + get packed with the base class). + + Reduces the size of FunctionExecutable from 160 to 152 bytes. + + * runtime/Executable.h: + (JSC::FunctionExecutable::variableCount): + (JSC::FunctionExecutable::FunctionExecutable): + +2010-05-15 Oliver Hunt <oliver@apple.com> + + Reviewed by Maciej Stachowiak. + + Incorrect codegen for slowcase of < in 64-bit + https://bugs.webkit.org/show_bug.cgi?id=39151 + + Call the correct stud for the slowcases of the < operator. + + * jit/JITArithmetic.cpp: + (JSC::JIT::emitSlow_op_jnless): + (JSC::JIT::emitSlow_op_jless): + +2010-05-15 Leo Yang <leo.yang@torchmobile.com.cn> + + Reviewed by Darin Adler. + + Fix bug https://bugs.webkit.org/show_bug.cgi?id=38890 + + * runtime/DatePrototype.cpp: + (JSC::formatLocaleDate): + +2010-05-15 Leandro Pereira <leandro@profusion.mobi> + + Reviewed by Adam Treat. + + [EFL] Add build system for the EFL port. + http://webkit.org/b/37945 + + * CMakeLists.txt: Added. + * CMakeListsEfl.txt: Added. + * config.h: Add conditional to look for cmakeconfig.h. + * jsc/CMakeLists.txt: Added. + * jsc/CMakeListsEfl.txt: Added. + * wtf/CMakeLists.txt: Added. + * wtf/CMakeListsEfl.txt: Added. + +2010-05-15 Chao-ying Fu <fu@mips.com> + + Reviewed by Oliver Hunt. + + Update MIPS JIT for unsigned right shift, Math.sqrt, load16 + https://bugs.webkit.org/show_bug.cgi?id=38412 + + Fixed MIPS build failure. + + * assembler/MIPSAssembler.h: + (JSC::MIPSAssembler::srl): + (JSC::MIPSAssembler::srlv): + (JSC::MIPSAssembler::sqrtd): + * assembler/MacroAssemblerMIPS.h: + (JSC::MacroAssemblerMIPS::urshift32): + (JSC::MacroAssemblerMIPS::sqrtDouble): + (JSC::MacroAssemblerMIPS::load16): + (JSC::MacroAssemblerMIPS::supportsFloatingPointSqrt): + * jit/JSInterfaceJIT.h: + +2010-05-15 Ilya Tikhonovsky <loislo@chromium.org> + + Reviewed by Geoffrey Garen. + + WebInspector: JSC Should provide heap size info for Timeline panel. + https://bugs.webkit.org/show_bug.cgi?id=38420 + + * JavaScriptCore.exp: + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + * runtime/Collector.cpp: + (JSC::Heap::size): + * runtime/Collector.h: + +2010-05-15 Dave Tapuska <dtapuska@rim.com> + + Reviewed by Darin Adler. + + Add a new define for we need strict arena alignment. ARMv5 with + double word instructions set this value. + + * wtf/Platform.h: + +2010-05-14 Stephanie Lewis <slewis@apple.com> + + Rubber-stamped by Mark Rowe. + + Update order files. + + * JavaScriptCore.order: + +2010-05-14 Jedrzej Nowacki <jedrzej.nowacki@nokia.com> + + Reviewed by Kenneth Rohde Christiansen. + + Implementation of QtScript API that provide method to convert a value to an object. + + [Qt] QtScript is missing toObject API + https://bugs.webkit.org/show_bug.cgi?id=36594 + + * qt/api/qscriptengine.cpp: + (QScriptEngine::toObject): + * qt/api/qscriptengine.h: + * qt/api/qscriptvalue.cpp: + (QScriptValue::toObject): + * qt/api/qscriptvalue.h: + * qt/api/qscriptvalue_p.h: + (QScriptValuePrivate::toObject): + * qt/tests/qscriptengine/tst_qscriptengine.cpp: + (tst_QScriptEngine::toObject): + (tst_QScriptEngine::toObjectTwoEngines): + * qt/tests/qscriptvalue/tst_qscriptvalue.cpp: + (tst_QScriptValue::toObjectSimple): + * qt/tests/qscriptvalue/tst_qscriptvalue.h: + +2010-05-14 Chao-ying Fu <fu@mips.com> + + Reviewed by Oliver Hunt. + + String Indexing Failure on JSVALUE32 targets + https://bugs.webkit.org/show_bug.cgi?id=39034 + + Remove zero-extend/shift-right code on regT1, because we already have + it in emit_op_get_by_val(). + + * jit/JITPropertyAccess.cpp: + (JSC::JIT::stringGetByValStubGenerator): + +2010-05-10 Jocelyn Turcotte <jocelyn.turcotte@nokia.com> + + Reviewed by Alexey Proskuryakov. + + Fix a VS2010 assert in std::copy + https://bugs.webkit.org/show_bug.cgi?id=38630 + + The assert complains that the output iterator is null. + + * wtf/Vector.h: + (WTF::::operator): + +2010-05-13 Maciej Stachowiak <mjs@apple.com> + + No review, attempted build fix. + + Try to fix Tiger build with some gratuitous initialization of + seemingly uninitialized variables. + + * wtf/text/StringImpl.h: + (WebCore::StringImpl::tryCreateUninitialized): + +2010-05-13 Maciej Stachowiak <mjs@apple.com> + + Reviewed by Mark Rowe. + + Various JavaScript string optimizations + https://bugs.webkit.org/show_bug.cgi?id=39051 + + Approximately 1% SunSpider speedup. + + * runtime/ArrayPrototype.cpp: + (JSC::arrayProtoFuncJoin): Remove branches from the hot code path + by moving the first pass outside the loop, and duplicating the hot loop + to extract the loop-invariant branch. + * runtime/RegExp.cpp: + (JSC::RegExp::match): resize ovector to 0 instead of clearing to avoid + thrash in case of large matches. + * runtime/RegExpConstructor.h: + (JSC::RegExpConstructor::performMatch): Mark ALWAYS_INLINE to make the + compiler respect our authority. + * runtime/StringPrototype.cpp: + (JSC::jsSpliceSubstringsWithSeparators): Inline. + (JSC::stringProtoFuncSubstring): Rewrite boundary condition checks to + reduce the number of floating point comparisons and branches. + +2010-05-12 Gavin Barraclough <barraclough@apple.com> + + Reviewed by Oliver Hunt. + + https://bugs.webkit.org/show_bug.cgi?id=39039 + Provide support for separate bytecode/JIT code translations for call/construct usage + This will allow us to produce code generated specifically for use as a constructor, not for general function use. + + * bytecode/CodeBlock.cpp: + (JSC::CodeBlock::CodeBlock): + (JSC::CodeBlock::unlinkCallers): + * bytecode/CodeBlock.h: + (JSC::CodeBlock::getBytecodeIndex): + (JSC::CodeBlock::getJITCode): + (JSC::CodeBlock::executablePool): + (JSC::GlobalCodeBlock::GlobalCodeBlock): + (JSC::FunctionCodeBlock::FunctionCodeBlock): + * interpreter/Interpreter.cpp: + (JSC::Interpreter::executeCall): + (JSC::Interpreter::executeConstruct): + (JSC::Interpreter::prepareForRepeatCall): + (JSC::Interpreter::execute): + (JSC::Interpreter::privateExecute): + * interpreter/Interpreter.h: + * jit/JIT.cpp: + (JSC::JIT::unlinkCallOrConstruct): + (JSC::JIT::linkConstruct): + * jit/JIT.h: + * jit/JITCall.cpp: + (JSC::JIT::compileOpCall): + (JSC::JIT::compileOpCallSlowCase): + * jit/JITOpcodes.cpp: + (JSC::JIT::privateCompileCTIMachineTrampolines): + * jit/JITOpcodes32_64.cpp: + (JSC::JIT::privateCompileCTIMachineTrampolines): + * jit/JITPropertyAccess.cpp: + (JSC::JIT::emitSlow_op_get_by_val): + * jit/JITPropertyAccess32_64.cpp: + (JSC::JIT::emitSlow_op_get_by_val): + * jit/JITStubs.cpp: + (JSC::DEFINE_STUB_FUNCTION): + * jit/JITStubs.h: + (JSC::JITThunks::ctiVirtualConstructLink): + (JSC::JITThunks::ctiVirtualConstruct): + (JSC::): + * jit/SpecializedThunkJIT.h: + (JSC::SpecializedThunkJIT::finalize): + * runtime/Arguments.h: + (JSC::JSActivation::copyRegisters): + * runtime/ArrayPrototype.cpp: + (JSC::isNumericCompareFunction): + * runtime/Executable.cpp: + (JSC::FunctionExecutable::~FunctionExecutable): + (JSC::FunctionExecutable::compileForCall): + (JSC::FunctionExecutable::compileForConstruct): + (JSC::EvalExecutable::generateJITCode): + (JSC::ProgramExecutable::generateJITCode): + (JSC::FunctionExecutable::generateJITCodeForCall): + (JSC::FunctionExecutable::generateJITCodeForConstruct): + (JSC::FunctionExecutable::markAggregate): + (JSC::FunctionExecutable::reparseExceptionInfo): + (JSC::EvalExecutable::reparseExceptionInfo): + (JSC::FunctionExecutable::recompile): + * runtime/Executable.h: + (JSC::ExecutableBase::ExecutableBase): + (JSC::ExecutableBase::isHostFunction): + (JSC::ExecutableBase::generatedJITCodeForCall): + (JSC::ExecutableBase::generatedJITCodeForConstruct): + (JSC::NativeExecutable::NativeExecutable): + (JSC::EvalExecutable::jitCode): + (JSC::ProgramExecutable::jitCode): + (JSC::FunctionExecutable::bytecodeForCall): + (JSC::FunctionExecutable::isGeneratedForCall): + (JSC::FunctionExecutable::generatedBytecodeForCall): + (JSC::FunctionExecutable::bytecodeForConstruct): + (JSC::FunctionExecutable::isGeneratedForConstruct): + (JSC::FunctionExecutable::generatedBytecodeForConstruct): + (JSC::FunctionExecutable::symbolTable): + (JSC::FunctionExecutable::FunctionExecutable): + (JSC::FunctionExecutable::jitCodeForCall): + (JSC::FunctionExecutable::jitCodeForConstruct): + * runtime/JSActivation.h: + (JSC::JSActivation::JSActivationData::JSActivationData): + * runtime/JSFunction.cpp: + (JSC::JSFunction::~JSFunction): + (JSC::JSFunction::call): + (JSC::JSFunction::construct): + * runtime/JSGlobalData.cpp: + (JSC::JSGlobalData::numericCompareFunction): + +2010-05-12 Oliver Hunt <oliver@apple.com> + + build fix + + <rdar://problem/7977960> REGRESSION (r59282): Tiger PPC build broken + + * runtime/UString.h: + (JSC::tryMakeString): + +2010-05-12 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + <rdar://7969718> Incorrect 32-bit codegen for op_jnlesseq + https://bugs.webkit.org/show_bug.cgi?id=39038 + + Correct the use of incorrect comparison operators. + + * jit/JITArithmetic32_64.cpp: + (JSC::JIT::emit_op_jlesseq): + (JSC::JIT::emitBinaryDoubleOp): + +2010-05-12 Caio Marcelo de Oliveira Filho <caio.oliveira@openbossa.org> + + Reviewed by Eric Seidel. + + JavaScriptCore: fix build when some optimizations are disabled + https://bugs.webkit.org/show_bug.cgi?id=38800 + + Remove outdated UNUSED_PARAMs when using JIT but with + ENABLE_JIT_OPTIMIZE_CALL and ENABLE_JIT_OPTIMIZE_PROPERTY_ACCESS + disabled. + + * jit/JITOpcodes.cpp: + * jit/JITOpcodes32_64.cpp: + +2010-05-12 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + Work around GCC stupidity by either explicitly inlining or by + using ALWAYS_INLINE + + * runtime/StringPrototype.cpp: + (JSC::stringProtoFuncReplace): + * runtime/UString.h: + (JSC::UString::~UString): + * wtf/RefPtr.h: + (WTF::RefPtr::RefPtr): + (WTF::RefPtr::~RefPtr): + * wtf/text/StringImpl.h: + (WebCore::StringImpl::create): + (WebCore::StringImpl::tryCreateUninitialized): + +2010-05-12 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Kenneth Rohde Christiansen. + + [Qt] Detect debug mode consistently + https://bugs.webkit.org/show_bug.cgi?id=38863 + + * JavaScriptCore.pri: + * JavaScriptCore.pro: + * jsc.pro: + * qt/api/QtScript.pro: + +2010-05-12 Peter Varga <pvarga@inf.u-szeged.hu> + + Reviewed by Gavin Barraclough. + + Remove a direct jump in generatePatternCharacterGreedy() and + generatePatternCharacterClassGreedy() functions if quantityCount of the term + is unlimited. + https://bugs.webkit.org/show_bug.cgi?id=38898 + + * yarr/RegexJIT.cpp: + (JSC::Yarr::RegexGenerator::generatePatternCharacterGreedy): + (JSC::Yarr::RegexGenerator::generateCharacterClassGreedy): + +2010-05-11 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Gavin Barraclough. + + Slight simplification to the calling convention: read RegisterFile::end() + from an absolute address, instead of from a pointer to the RegisterFile + on the stack. + + SunSpider reports no change. + + This should free us to remove the RegisterFile argument from the stack. + + * jit/JIT.cpp: + (JSC::JIT::privateCompile): + +2010-05-11 Brian Weinstein <bweinstein@apple.com> + + Possible Gtk build fix - use ALWAYS_INLINE on functions that were inlined. + + * wtf/text/AtomicString.h: + (WebCore::AtomicString::add): Use ALWAYS_INLINE. + * wtf/text/WTFString.h: + (WebCore::String::length): Ditto. + (WebCore::String::isEmpty): Ditto. + +2010-05-11 Anders Carlsson <andersca@apple.com> + + Reviewed by Sam Weinig. + + Add a compile time assert that sizeof(String) == sizeof(AtomicString). + + * wtf/text/AtomicString.cpp: + +2010-05-11 Anders Carlsson <andersca@apple.com> + + Reviewed by Mark Rowe. + + https://bugs.webkit.org/show_bug.cgi?id=38961 + Move some member variables around to reduce class sizes. + + Make AtomicString no longer inherit from FastAllocBase. + + Since AtomicString's first (and only) member variable, a RefPtr, also inherits from FastAllocBase this + was causing the size of AtomicString to contain sizeof(void*) bytes of padding. + + * wtf/text/AtomicString.h: + +2010-05-11 Brian Weinstein <bweinstein@apple.com> + + Reviewed by Maciej Stachowiak. + Patch by Gavin Barraclough. + + Rest of REGRESSION (r57900-57919): 3% PLT Regression from moving strings into WTF. + https://bugs.webkit.org/show_bug.cgi?id=38930 + <rdar://problem/7937188> + + Inline string functions that are commonly used in WTFString, CString, and AtomicString. + + * JavaScriptCore.exp: Remove exports of inline functions. + * wtf/text/AtomicString.cpp: + (WebCore::AtomicString::addSlowCase): Rename add to addSlowCase, and inline the fast path + of add. + * wtf/text/AtomicString.h: + (WebCore::AtomicString::add): Inline the fast path of add. + * wtf/text/CString.cpp: Inline implementation of data and length in the header. + * wtf/text/CString.h: + (WTF::CString::data): Inline implementation of data. + (WTF::CString::length): Inline implementation of length. + * wtf/text/WTFString.cpp: Remove implementations of functions to be inlined. + * wtf/text/WTFString.h: + (WebCore::String::String): Inline implementation. + (WebCore::String::length): Ditto. + (WebCore::String::characters): Ditto. + (WebCore::String::operator[]): Ditto. + (WebCore::String::isEmpty): Ditto. + +2010-05-11 Brian Weinstein <bweinstein@apple.com> + + Reviewed by Steve Falkenburg and Jon Honeycutt. + Patch by Gavin Barraclough. + + REGRESSION (r57900-57919): 3% PLT Regression from moving strings into WTF. + https://bugs.webkit.org/show_bug.cgi?id=38930 + <rdar://problem/7937188> + + When the String classes were moved from WebCore to WTF, it meant that on Windows, all operations + on Strings in WebCore had to cross a DLL boundary (from WebKit.dll to JavaScript.dll). + + We fix this by refactoring some of the WTF string code, so the code in AtomicString, StringImpl, and + WTFString can be built by both WebCore and WTF, and we don't need to talk across a DLL to do operations + on Strings. + + * GNUmakefile.am: Add new file to build system. + * JavaScriptCore.gypi: Ditto. + * JavaScriptCore.pro: Ditto. + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: Remove string exports, because these are now + handled in WebCore. + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.vcproj: Remove the post-build step that was added + here (the post build step is in JavaScriptCoreCommon.vsprops). + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreCommon.vsprops: Copy the three cpp files that need to be compiled + by WebCore into the WebKitOutputDir directory. + * JavaScriptCore.vcproj/WTF/WTF.vcproj: Add the StringStatics file. + * JavaScriptCore.vcproj/jsc/jsc.vcproj: Add the three WTF string cpp files to this project. + * JavaScriptCore.vcproj/jsc/jscCommon.vsprops: Remove the need to link against WTF.lib (since jsc links against JavaScriptCore). + * JavaScriptCore.xcodeproj/project.pbxproj: Add the StringStatics file. + * wtf/text/AtomicString.cpp: Moved code to StringStatics. + * wtf/text/StringImpl.cpp: Ditto. + * wtf/text/StringStatics.cpp: Added. Move functions in WTF Strings that define static variables to here, so + the rest of the files can be compiled in WebCore. + (WebCore::StringImpl::empty): Moved from StringImpl.cpp to here. + (WebCore::AtomicString::init): Moved from AtomicString.cpp to here. + +2010-05-11 Alice Liu <alice.liu@apple.com> + + Rubber-stamped by Gavin Barraclough. + + Fix build error when enabling debugging block in WebKit win painting code + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + +2010-05-11 Mark Rowe <mrowe@apple.com> + + Fix the world. + + In r59162 a change was made to WebCore's FeatureDefines.xcconfig that enabled FILE_READER and FILE_WRITER. + The author and reviewer of that patch ignored the carefully-worded warning at the top of that file asking + that changes to the file be kept in sync across JavaScriptCore, WebCore and WebKit, as well as being kept + in sync with build-webkit. This led to WebCore and WebKit having different views of Document's vtable + and results in crashes in Safari shortly after launch when virtual function calls resulted in the wrong + function in WebCore being called. + + We fix this by bringing the FeatureDefines.xcconfig files in to sync. Based on the ChangeLog message and + other changes in r59162 it appears that enabling FILE_WRITER was unintentional so that particular change + has been reverted. + + * Configurations/FeatureDefines.xcconfig: + +2010-05-11 Sheriff Bot <webkit.review.bot@gmail.com> + + Unreviewed, rolling out r59171. + http://trac.webkit.org/changeset/59171 + https://bugs.webkit.org/show_bug.cgi?id=38933 + + "Broke the world" (Requested by bweinstein on #webkit). + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.vcproj: + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreCommon.vsprops: + * JavaScriptCore.vcproj/WTF/WTF.vcproj: + * JavaScriptCore.vcproj/jsc/jsc.vcproj: + * JavaScriptCore.vcproj/jsc/jscCommon.vsprops: + * JavaScriptCore.xcodeproj/project.pbxproj: + * wtf/text/AtomicString.cpp: + (WebCore::AtomicString::init): + * wtf/text/StringImpl.cpp: + (WebCore::StringImpl::empty): + * wtf/text/StringStatics.cpp: Removed. + +2010-05-11 Brian Weinstein <bweinstein@apple.com> + + Reviewed by Steve Falkenburg. + Patch by Gavin Barraclough. + + REGRESSION (r57900-57919): 3% PLT Regression from moving strings into WTF. + https://bugs.webkit.org/show_bug.cgi?id=38930 + <rdar://problem/7937188> + + When the String classes were moved from WebCore to WTF, it meant that on Windows, all operations + on Strings in WebCore had to cross a DLL boundary (from WebKit.dll to JavaScript.dll). + + We fix this by refactoring some of the WTF string code, so the code in AtomicString, StringImpl, and + WTFString can be built by both WebCore and WTF, and we don't need to talk across a DLL to do operations + on Strings. + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: Remove string exports, because these are now + handled in WebCore. + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.vcproj: Remove the post-build step that was added + here (the post build step is in JavaScriptCoreCommon.vsprops). + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreCommon.vsprops: Copy the three cpp files that need to be compiled + by WebCore into the WebKitOutputDir directory. + * JavaScriptCore.vcproj/WTF/WTF.vcproj: Add the StringStatics file. + * JavaScriptCore.vcproj/jsc/jsc.vcproj: Add the three WTF string cpp files to this project. + * JavaScriptCore.vcproj/jsc/jscCommon.vsprops: Remove the need to link against WTF.lib (since jsc links against JavaScriptCore). + * JavaScriptCore.xcodeproj/project.pbxproj: Add the StringStatics file. + * wtf/text/AtomicString.cpp: Moved code to StringStatics. + * wtf/text/StringImpl.cpp: Ditto. + * wtf/text/StringStatics.cpp: Added. Move functions in WTF Strings that define static variables to here, so + the rest of the files can be compiled in WebCore. + (WebCore::StringImpl::empty): Moved from StringImpl.cpp to here. + (WebCore::AtomicString::init): Moved from AtomicString.cpp to here. + +2010-05-11 Geoffrey Garen <ggaren@apple.com> + + Fixed test failures seen on SnowLeopard buildbot. + + * runtime/JSString.cpp: + (JSC::JSString::replaceCharacter): Don't use size_t and wtf::notFound. + Instead, use the individual types and notFound values of the string APIs + we're using, since they're not necessarily the same in 64bit. + +2010-05-11 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Oliver Hunt and Darin Adler. + + Start using ropes in String.prototype.replace. + + 1%-1.5% speedup on SunSpider. + + * runtime/JSString.cpp: + (JSC::JSString::resolveRope): Updated for RopeImpl refactoring. + + (JSC::JSString::replaceCharacter): Added a replaceCharacter function, which creates + a rope for the resulting replacement. + + * runtime/JSString.h: A few changes here: + (JSC::): + (JSC::RopeBuilder::RopeIterator::RopeIterator): + (JSC::RopeBuilder::RopeIterator::operator++): + (JSC::RopeBuilder::RopeIterator::operator*): + (JSC::RopeBuilder::RopeIterator::operator!=): + (JSC::RopeBuilder::RopeIterator::WorkItem::WorkItem): + (JSC::RopeBuilder::RopeIterator::WorkItem::operator!=): + (JSC::RopeBuilder::RopeIterator::skipRopes): Created a RopeIterator abstraction. + We use this to do a substring find without having to resolve the rope. + (We could use this iterator when resolving ropes, too, but resolving + ropes backwards is usually more efficient.) + + (JSC::RopeBuilder::JSString): Added constructors for 2 & 3 UStrings. + + (JSC::RopeBuilder::appendValueInConstructAndIncrementLength): + (JSC::RopeBuilder::size): Updated for RopeImpl refactoring. + + * runtime/Operations.h: Updated for RopeImpl refactoring. + (JSC::jsString): Added jsString functions for 2 & 3 UStrings. + + * runtime/RopeImpl.cpp: + (JSC::RopeImpl::derefFibersNonRecursive): + * runtime/RopeImpl.h: + (JSC::RopeImpl::initializeFiber): + (JSC::RopeImpl::size): + (JSC::RopeImpl::fibers): + (JSC::RopeImpl::deref): + (JSC::RopeImpl::RopeImpl): A little refactoring to make this patch easier: + Moved statics to the top of the class; put multi-statement functions on + multiple lines; renamed "fiberCount" to "size" to match other collections; + changed the "fibers" accessor to return the fibers buffer, instead of an + item in the buffer, to make iteration easier. + + * runtime/StringPrototype.cpp: + (JSC::stringProtoFuncReplace): Don't resolve a rope unless we need to. Do + use our new replaceCharacter function if possible. Do use a rope to + represent splicing three strings together. + +2010-05-10 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Darin Adler. + + [Qt] Disable JIT support for mingw-w64 + https://bugs.webkit.org/show_bug.cgi?id=38747 + + Disale JIT for mingw-w64 as it is reportedly + unstable. + + Thanks for Vanboxem Rruben for the investigation. + + * wtf/Platform.h: + +2010-05-09 Fumitoshi Ukai <ukai@chromium.org> + + Reviewed by Eric Seidel. + + JavaScriptCore/wtf/MD5.h: checksum should take a reference to output. + https://bugs.webkit.org/show_bug.cgi?id=38723 + + * JavaScriptCore.exp: + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + * wtf/MD5.cpp: + (WTF::expectMD5): + Fix for checksum change. + (WTF::MD5::checksum): + Take a reference to output, instead of returning the result by value, to reduce coping for performance. + * wtf/MD5.h: + +2010-05-09 Oliver Hunt <oliver@apple.com> + + Build fix. + + Fix accidental repeat addition of emit_op_new_regexp + + * jit/JITOpcodes.cpp: + +2010-05-09 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + REGRESSION(r57955): RegExp literals should not actually be cached, so r57955 should be rolled out. + https://bugs.webkit.org/show_bug.cgi?id=38828 + <rdar://problem/7961634> + + Rollout r57955 + + * bytecode/CodeBlock.cpp: + (JSC::regexpToSourceString): + (JSC::regexpName): + (JSC::CodeBlock::dump): + (JSC::CodeBlock::shrinkToFit): + * bytecode/CodeBlock.h: + (JSC::CodeBlock::addRegExp): + (JSC::CodeBlock::regexp): + * bytecode/Opcode.h: + * bytecompiler/BytecodeGenerator.cpp: + (JSC::BytecodeGenerator::addRegExp): + (JSC::BytecodeGenerator::emitNewRegExp): + * bytecompiler/BytecodeGenerator.h: + * bytecompiler/NodesCodegen.cpp: + (JSC::RegExpNode::emitBytecode): + * interpreter/Interpreter.cpp: + (JSC::Interpreter::privateExecute): + * jit/JIT.cpp: + (JSC::JIT::privateCompileMainPass): + * jit/JIT.h: + * jit/JITOpcodes.cpp: + (JSC::JIT::emit_op_new_regexp): + * jit/JITStubs.cpp: + (JSC::DEFINE_STUB_FUNCTION): + * jit/JITStubs.h: + (JSC::): + +2010-05-09 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Oliver Hunt. + + Reserve a large-ish initial capacity for Lexer::m_buffer16. + + SunSpider says 0.3% faster. + + m_buffer16 is used when parsing complex strings -- for example, strings + with escape sequences in them. These kinds of strings can be really long, + and we want to avoid repeatedly copying as we grow m_buffer16. + + The net memory cost is quite low, since it's proporitional to source + code we already have in memory, and we throw away m_buffer16 right when + we're done parsing. + + * parser/Lexer.cpp: + (JSC::Lexer::Lexer): No need to reserve initial capacity in our constructor, + since setCode will be called before we're asked to lex anything. + (JSC::Lexer::setCode): Reserve enough space to lex half the source code + as a complex string without having to copy. + (JSC::Lexer::clear): No need to reserve initial capacity here either, + since setCode will be called before we're asked to lex anything. + +2010-05-09 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Eric Seidel. + + [Qt] Remove YARR configuration rules from Qt make system + https://bugs.webkit.org/show_bug.cgi?id=38819 + + Setting YARR based on JIT seeting is now in Platform.h + for all ports. This patch essentially reverses r49238. + + * JavaScriptCore.pri: + +2010-05-09 Oliver Hunt <oliver@apple.com> + + Reviewed by Maciej Stachowiak. + + Improve string indexing performance + https://bugs.webkit.org/show_bug.cgi?id=38814 + + Add an assembly stub to do indexed loads from strings much + more cheaply than the current stub dispatch logic. We can + do this because we are able to make guarantees about the + register contents when entering the stub so the call overhead + is negligible. + + * jit/JIT.h: + * jit/JITInlineMethods.h: + * jit/JITOpcodes.cpp: + * jit/JITPropertyAccess.cpp: + (JSC::JIT::stringGetByValStubGenerator): + (JSC::JIT::emitSlow_op_get_by_val): + Moved from JITOpcodes.cpp to keep the slowcase next to + the normal case codegen as we do for everything else. + * jit/JITPropertyAccess32_64.cpp: + (JSC::JIT::stringGetByValStubGenerator): + (JSC::JIT::emitSlow_op_get_by_val): + * jit/JSInterfaceJIT.h: + (JSC::JSInterfaceJIT::emitFastArithImmToInt): + +2010-05-09 Maciej Stachowiak <mjs@apple.com> + + Fixed version of: "Optimized o[s] where o is a cell and s is a string" + https://bugs.webkit.org/show_bug.cgi?id=38815 + + Fixed the previous patch for this from Geoff Garen. + + The two problems were a missing exception check and a PropertySlot + initialized improperly, leading to crashes and failures in the case + of getters accessed with bracket syntax. + + Previous patch: + + Optimized o[s] where o is a cell and s is a string, removing some old + code that wasn't really tuned for the JIT. + + SunSpider says 0.8% faster. + + * jit/JITStubs.cpp: + (JSC::DEFINE_STUB_FUNCTION): + * runtime/JSCell.h: + +2010-05-08 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Unreviewed, build fix. + + Fix missing terminating ' character warning + + The warning has been only seen on the Gtk buildbots. + + * wtf/Platform.h: + +2010-05-08 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Gavin Barraclough. + + Refactor YARR and YARR_JIT default rules in Platform.h + https://bugs.webkit.org/show_bug.cgi?id=38727 + + Turn on YARR and YARR JIT by default of JIT is enabled. + Ports can overrule this default decisions as they wish. + + * wtf/Platform.h: + +2010-05-08 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + Split JSVALUE32_64 code out of JITOpcodes.cpp and into JITOpcodes32_64.cpp + https://bugs.webkit.org/show_bug.cgi?id=38808 + + * GNUmakefile.am: + * JavaScriptCore.gypi: + * JavaScriptCore.pro: + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.vcproj: + * JavaScriptCore.xcodeproj/project.pbxproj: + * jit/JITOpcodes.cpp: + * jit/JITOpcodes32_64.cpp: Added. + (JSC::JIT::privateCompileCTIMachineTrampolines): + (JSC::JIT::emit_op_mov): + (JSC::JIT::emit_op_end): + (JSC::JIT::emit_op_jmp): + (JSC::JIT::emit_op_loop_if_lesseq): + (JSC::JIT::emitSlow_op_loop_if_lesseq): + (JSC::JIT::emit_op_new_object): + (JSC::JIT::emit_op_instanceof): + (JSC::JIT::emitSlow_op_instanceof): + (JSC::JIT::emit_op_new_func): + (JSC::JIT::emit_op_get_global_var): + (JSC::JIT::emit_op_put_global_var): + (JSC::JIT::emit_op_get_scoped_var): + (JSC::JIT::emit_op_put_scoped_var): + (JSC::JIT::emit_op_tear_off_activation): + (JSC::JIT::emit_op_tear_off_arguments): + (JSC::JIT::emit_op_new_array): + (JSC::JIT::emit_op_resolve): + (JSC::JIT::emit_op_to_primitive): + (JSC::JIT::emitSlow_op_to_primitive): + (JSC::JIT::emit_op_strcat): + (JSC::JIT::emit_op_resolve_base): + (JSC::JIT::emit_op_resolve_skip): + (JSC::JIT::emit_op_resolve_global): + (JSC::JIT::emitSlow_op_resolve_global): + (JSC::JIT::emit_op_not): + (JSC::JIT::emitSlow_op_not): + (JSC::JIT::emit_op_jfalse): + (JSC::JIT::emitSlow_op_jfalse): + (JSC::JIT::emit_op_jtrue): + (JSC::JIT::emitSlow_op_jtrue): + (JSC::JIT::emit_op_jeq_null): + (JSC::JIT::emit_op_jneq_null): + (JSC::JIT::emit_op_jneq_ptr): + (JSC::JIT::emit_op_jsr): + (JSC::JIT::emit_op_sret): + (JSC::JIT::emit_op_eq): + (JSC::JIT::emitSlow_op_eq): + (JSC::JIT::emit_op_neq): + (JSC::JIT::emitSlow_op_neq): + (JSC::JIT::compileOpStrictEq): + (JSC::JIT::emit_op_stricteq): + (JSC::JIT::emitSlow_op_stricteq): + (JSC::JIT::emit_op_nstricteq): + (JSC::JIT::emitSlow_op_nstricteq): + (JSC::JIT::emit_op_eq_null): + (JSC::JIT::emit_op_neq_null): + (JSC::JIT::emit_op_resolve_with_base): + (JSC::JIT::emit_op_new_func_exp): + (JSC::JIT::emit_op_throw): + (JSC::JIT::emit_op_get_pnames): + (JSC::JIT::emit_op_next_pname): + (JSC::JIT::emit_op_push_scope): + (JSC::JIT::emit_op_pop_scope): + (JSC::JIT::emit_op_to_jsnumber): + (JSC::JIT::emitSlow_op_to_jsnumber): + (JSC::JIT::emit_op_push_new_scope): + (JSC::JIT::emit_op_catch): + (JSC::JIT::emit_op_jmp_scopes): + (JSC::JIT::emit_op_switch_imm): + (JSC::JIT::emit_op_switch_char): + (JSC::JIT::emit_op_switch_string): + (JSC::JIT::emit_op_new_error): + (JSC::JIT::emit_op_debug): + (JSC::JIT::emit_op_enter): + (JSC::JIT::emit_op_enter_with_activation): + (JSC::JIT::emit_op_create_arguments): + (JSC::JIT::emit_op_init_arguments): + (JSC::JIT::emit_op_convert_this): + (JSC::JIT::emitSlow_op_convert_this): + (JSC::JIT::emit_op_profile_will_call): + (JSC::JIT::emit_op_profile_did_call): + +2010-05-08 Gabor Loki <loki@webkit.org> + + Reviewed by Gavin Barraclough. + + Fix halfword loads on ARM + https://bugs.webkit.org/show_bug.cgi?id=38741 + + The BaseIndex and ImplicitAddress are contain 32bit wide offset, but + the load16 functions were working with 8 bit data (encoded in the + instruction). If the offset cannot be encoded in an instruction, it + should be stored in a temporary register. + + * assembler/ARMAssembler.h: + (JSC::ARMAssembler::getOffsetForHalfwordDataTransfer): + * assembler/MacroAssemblerARM.h: + (JSC::MacroAssemblerARM::load16): + +2010-05-08 Gabor Loki <loki@webkit.org> + + Reviewed by Gavin Barraclough. + + Fix spanning branch instruction on Cortex-A8 with Thumb-2 JIT + https://bugs.webkit.org/show_bug.cgi?id=38280 + + If the 32-bit Thumb-2 branch instruction spans two 4KiB regions and + the target of the branch falls within the first region it is + possible for the processor to incorrectly determine the branch + instruction, and it is also possible in some cases for the processor + to enter a deadlock state. + + * assembler/ARMv7Assembler.h: + (JSC::ARMv7Assembler::linkJumpAbsolute): + +2010-05-08 Adam Barth <abarth@webkit.org> + + Reviewed by Darin Fisher. + + [WTFURL] Add a buffer for managing the output of the canonicalizer + https://bugs.webkit.org/show_bug.cgi?id=38671 + + The canonicalizer produces a stream of canonicalized characters, which + we will store in a URLBuffer. URLBuffer is quite similar to Vector, + but serves as an abstraction to isolate us from the underlying + character storage. In the case of WebKit, we'll probably implement a + subclass of URLBuffer with Vector. In Chromium, the backing store is a + std::string, which avoids an unnecessary memcpy that would be caused by + using Vector instead. + + * wtf/url/src/URLBuffer.h: Added. + (WTF::URLBuffer::URLBuffer): + (WTF::URLBuffer::~URLBuffer): + (WTF::URLBuffer::at): + (WTF::URLBuffer::set): + (WTF::URLBuffer::capacity): + (WTF::URLBuffer::length): + (WTF::URLBuffer::data): + (WTF::URLBuffer::setLength): + (WTF::URLBuffer::append): + (WTF::URLBuffer::grow): + +2010-05-08 Eric Seidel <eric@webkit.org> + + Unreviewed, just reverting commit. + + REGRESSION(59000): r59000 contained all sorts of changes it should not have, needs revert. + https://bugs.webkit.org/show_bug.cgi?id=38798 + + * bytecompiler/BytecodeGenerator.cpp: + (JSC::BytecodeGenerator::emitResolve): + * jit/JITOpcodes.cpp: + (JSC::JIT::emit_op_resolve_global): + +2010-05-04 Jer Noble <jer.noble@apple.com> + + Reviewed by Maciej Stachowiak + + Safari pegs CPU and drops tons of frames using HTML5 Vimeo player + <https://bugs.webkit.org/show_bug.cgi?id=34005> + <rdar://problem/7569713> + + Added cancelCallOnMainThread. callOnMainThread should always now be paired + with cancelCallOnMainThread in situations where the refcon passed to callOnMainThread + may be dealloced before the main thread function can be dispatched. + + * wtf/MainThread.cpp: + (WTF::FunctionWithContext::operator == ): Supports the FunctionWithContextFinder predicate functor. + (WTF::FunctionWithContextFinder::FunctionWithContextFinder): Predicate functor for use with Dequeue::findIf + (WTF::FunctionWithContextFinder::operator()): + (WTF::cancelCallOnMainThread): + * wtf/MainThread.h: + +2010-05-07 Oliver Hunt <oliver@apple.com> + + Windows build fix. + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + +2010-05-07 Eric Seidel <eric@webkit.org> + + Unreviewed, rolling out r58990. + http://trac.webkit.org/changeset/58990 + + Broke 4 tests on Snow Leopard and Qt. + + * jit/JITStubs.cpp: + (JSC::DEFINE_STUB_FUNCTION): + * runtime/JSCell.h: + +2010-05-07 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + Optimize access to the global object from a function that uses eval + https://bugs.webkit.org/show_bug.cgi?id=38644 + + Fix bug where cross scope access to a global var (vs. property) would + be allowed without checking for intervening dynamic scopes. + + * bytecompiler/BytecodeGenerator.cpp: + (JSC::BytecodeGenerator::emitResolve): + +2010-05-07 Oliver Hunt <oliver@apple.com> + + 32-bit buildfix. + + Macro expansion I stab at thee! + + * jit/JITOpcodes.cpp: + (JSC::JIT::emit_op_resolve_global): + +2010-05-07 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Darin Adler. + + Optimized o[s] where o is a cell and s is a string, removing some old + code that wasn't really tuned for the JIT. + + SunSpider says 0.8% faster. + + * jit/JITStubs.cpp: + (JSC::DEFINE_STUB_FUNCTION): + * runtime/JSCell.h: + +2010-05-07 Oliver Hunt <oliver@apple.com> + + Windows build fix. + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + +2010-05-07 Oliver Hunt <oliver@apple.com> + + Reviewed by Geoffrey Garen. + + Optimize access to the global object from a function that uses eval + https://bugs.webkit.org/show_bug.cgi?id=38644 + + Add op_resolve_global_dynamic, a variant of op_resolve_global that + checks each node in the scope chain for dynamically inserted properties + and falls back to the normal resolve logic in that case. + + * JavaScriptCore.exp: + * bytecode/CodeBlock.cpp: + (JSC::isGlobalResolve): + (JSC::CodeBlock::printStructures): + (JSC::CodeBlock::dump): + (JSC::CodeBlock::derefStructures): + * bytecode/Opcode.h: + * bytecompiler/BytecodeGenerator.cpp: + (JSC::BytecodeGenerator::findScopedProperty): + Now take an additional reference parameter to used to indicate that + there were nodes that may gain dynamic properties + (JSC::BytecodeGenerator::emitResolve): + (JSC::BytecodeGenerator::emitResolveBase): + (JSC::BytecodeGenerator::emitResolveWithBase): + deal with additional argument to findScopedProperty + * bytecompiler/BytecodeGenerator.h: + * bytecompiler/NodesCodegen.cpp: + (JSC::FunctionCallResolveNode::emitBytecode): + (JSC::PostfixResolveNode::emitBytecode): + (JSC::PrefixResolveNode::emitBytecode): + (JSC::ReadModifyResolveNode::emitBytecode): + (JSC::AssignResolveNode::emitBytecode): + These functions use findScopedProperty directly in order to + optimise lookup. They cannot trivially handle any degree of + dynamism in the lookup so we just give up in such case. + * interpreter/Interpreter.cpp: + (JSC::Interpreter::resolveGlobalDynamic): + (JSC::Interpreter::execute): + (JSC::Interpreter::privateExecute): + * interpreter/Interpreter.h: + * jit/JIT.cpp: + (JSC::JIT::privateCompileMainPass): + (JSC::JIT::privateCompileSlowCases): + * jit/JIT.h: + * jit/JITOpcodes.cpp: + (JSC::JIT::emit_op_resolve_global): + (JSC::JIT::emit_op_resolve_global_dynamic): + (JSC::JIT::emitSlow_op_resolve_global): + (JSC::JIT::emitSlow_op_resolve_global_dynamic): + Happily resolve_global_dynamic can share the slow case! + * jit/JITStubs.h: + (JSC::): + * runtime/JSActivation.cpp: + (JSC::JSActivation::isDynamicScope): + * runtime/JSActivation.h: + * runtime/JSGlobalObject.cpp: + (JSC::JSGlobalObject::isDynamicScope): + * runtime/JSGlobalObject.h: + * runtime/JSStaticScopeObject.cpp: + (JSC::JSStaticScopeObject::isDynamicScope): + * runtime/JSStaticScopeObject.h: + * runtime/JSVariableObject.h: + +2010-05-07 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Gavin Barraclough. + + Fixed Sputnik failure seen on buildbot. + + * runtime/NumberPrototype.cpp: + (JSC::numberProtoFuncToString): Cast to unsigned before checking magnitude, + to efficiently exclude negative numbers, which require two characters + instead of one. + +2010-05-07 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Alexey Proskuryakov. + + Slightly more effective way to guarantee a compile-time constant, at + least on gcc. + + * runtime/NumberPrototype.cpp: + (JSC::numberProtoFuncToString): Use pointer notation instead of array + notation, so the compiler treats the string and its contents as constant. + +2010-05-07 Adam Barth <abarth@webkit.org> + + Reviewed by Alexey Proskuryakov. + + [WTFURL] Add core URL parser + https://bugs.webkit.org/show_bug.cgi?id=38572 + + This patch adds the core of the URL parser. The URL parser uses a + templated notion of a code unit to support different string types. + Later we'll add some non-templated APIs with concrete types that + clients can use more conveniently. + + The URLParser has a couple calls to ASSERT(), which I've commented out + for now. I'm not 100% sure how to handle the dependency issues there + yet, so I've punted on the issue for now. + + Also, there are a number of methods that are declared public in this + patch that are only used by the canonicalizer. My plan is to make + those private or protected and make the canonicalizer a friend of the + parser. The details will be in a later patch. + + * wtf/url/src/URLComponent.h: + (WTF::URLComponent::fromRange): + (WTF::URLComponent::isNonEmpty): + (WTF::URLComponent::isEmptyOrInvalid): + * wtf/url/src/URLParser.h: Added. + (WTF::URLParser::): + (WTF::URLParser::isPossibleAuthorityTerminator): + (WTF::URLParser::parseAuthority): + (WTF::URLParser::extractScheme): + (WTF::URLParser::parseAfterScheme): + (WTF::URLParser::parseStandardURL): + (WTF::URLParser::parsePath): + (WTF::URLParser::parsePathURL): + (WTF::URLParser::parseMailtoURL): + (WTF::URLParser::parsePort): + (WTF::URLParser::extractFileName): + (WTF::URLParser::extractQueryKeyValue): + (WTF::URLParser::isURLSlash): + (WTF::URLParser::shouldTrimFromURL): + (WTF::URLParser::trimURL): + (WTF::URLParser::consecutiveSlashes): + (WTF::URLParser::isPortDigit): + (WTF::URLParser::nextAuthorityTerminator): + (WTF::URLParser::parseUserInfo): + (WTF::URLParser::parseServerInfo): + +2010-05-07 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Oliver Hunt. + + Added a fast path for number-to-character conversion via + Number.prototype.toString base 36. + + 0.7% speedup on SunSpider. + + * runtime/NumberPrototype.cpp: + (JSC::numberProtoFuncToString): Made radix interpretation a little more + efficient by keeping it in int space. Turned "const char" into + "static const char" just in case. Added a fast path for base 36 + conversion of something that will turn into a character. + +2010-05-07 Adam Barth <abarth@webkit.org> + + Reviewed by Darin Fisher. + + [WTFURL] Add a malloc-backed URLBuffer + https://bugs.webkit.org/show_bug.cgi?id=38694 + + This patch adds a URLBuffer that uses new/delete to manage the backing + store for the buffer. + + * wtf/url/src/RawURLBuffer.h: Added. + (WTF::RawURLBuffer::RawURLBuffer): + (WTF::RawURLBuffer::~RawURLBuffer): + (WTF::RawURLBuffer::resize): + 2010-05-06 Fumitoshi Ukai <ukai@chromium.org> Reviewed by Alexey Proskuryakov. |