diff options
Diffstat (limited to 'JavaScriptCore/ChangeLog')
| -rw-r--r-- | JavaScriptCore/ChangeLog | 2452 |
1 files changed, 2452 insertions, 0 deletions
diff --git a/JavaScriptCore/ChangeLog b/JavaScriptCore/ChangeLog index fb09372..3e9187b 100644 --- a/JavaScriptCore/ChangeLog +++ b/JavaScriptCore/ChangeLog @@ -1,3 +1,2455 @@ +2009-12-10 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + Incorrect caching of prototype lookup with dictionary base + https://bugs.webkit.org/show_bug.cgi?id=32402 + + Make sure we don't add cached prototype lookup to the proto_list + lookup chain if the top level object is a dictionary. + + * jit/JITStubs.cpp: + (JSC::JITThunks::tryCacheGetByID): + +2009-12-10 Gavin Barraclough <barraclough@apple.com> + + Reviewed by Oliver Hunt. + + https://bugs.webkit.org/show_bug.cgi?id=32400 + Switch remaining cases of string addition to use ropes. + + ~1% progression on Sunspidey. + + * jit/JITStubs.cpp: + (JSC::DEFINE_STUB_FUNCTION): + * runtime/JSString.h: + (JSC::JSString::JSString): + (JSC::JSString::appendStringInConstruct): + * runtime/Operations.cpp: + (JSC::jsAddSlowCase): + * runtime/Operations.h: + (JSC::jsString): + (JSC::jsAdd): + +2009-12-10 Kent Hansen <kent.hansen@nokia.com> + + Reviewed by Geoffrey Garen. + + Remove JSObject::getPropertyAttributes() and all usage of it. + https://bugs.webkit.org/show_bug.cgi?id=31933 + + getOwnPropertyDescriptor() should be used instead. + + * JavaScriptCore.exp: + * JavaScriptCore.order: + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + * debugger/DebuggerActivation.cpp: + (JSC::DebuggerActivation::getOwnPropertyDescriptor): + * debugger/DebuggerActivation.h: + * runtime/JSObject.cpp: + (JSC::JSObject::propertyIsEnumerable): + * runtime/JSObject.h: + * runtime/JSVariableObject.cpp: + * runtime/JSVariableObject.h: + +2009-12-10 Gavin Barraclough <barraclough@apple.com> + + Reviewed by Oliver Hunt & Mark Rowe. + + https://bugs.webkit.org/show_bug.cgi?id=32367 + Add support for short Ropes (up to 3 entries) inline within JSString. + (rather than externally allocating an object to hold the rope). + Switch jsAdd of (JSString* + JSString*) to now make use of Ropes. + + ~1% progression on Sunspidey. + + * interpreter/Interpreter.cpp: + (JSC::Interpreter::privateExecute): + * jit/JITOpcodes.cpp: + (JSC::JIT::privateCompileCTIMachineTrampolines): + * jit/JITStubs.cpp: + (JSC::DEFINE_STUB_FUNCTION): + * runtime/JSString.cpp: + (JSC::JSString::resolveRope): + (JSC::JSString::toBoolean): + (JSC::JSString::getStringPropertyDescriptor): + * runtime/JSString.h: + (JSC::JSString::Rope::Fiber::deref): + (JSC::JSString::Rope::Fiber::ref): + (JSC::JSString::Rope::Fiber::refAndGetLength): + (JSC::JSString::Rope::append): + (JSC::JSString::JSString): + (JSC::JSString::~JSString): + (JSC::JSString::value): + (JSC::JSString::tryGetValue): + (JSC::JSString::length): + (JSC::JSString::canGetIndex): + (JSC::JSString::appendStringInConstruct): + (JSC::JSString::appendValueInConstructAndIncrementLength): + (JSC::JSString::isRope): + (JSC::JSString::string): + (JSC::JSString::ropeLength): + (JSC::JSString::getStringPropertySlot): + * runtime/Operations.h: + (JSC::jsString): + (JSC::jsAdd): + (JSC::resolveBase): + +2009-12-09 Anders Carlsson <andersca@apple.com> + + Reviewed by Geoffrey Garen. + + Fix three more things found by compiling with clang++. + + * runtime/Structure.h: + (JSC::StructureTransitionTable::reifySingleTransition): + Add the 'std' qualifier to the call to make_pair. + + * wtf/DateMath.cpp: + (WTF::initializeDates): + Incrementing a bool is deprecated according to the C++ specification. + + * wtf/PtrAndFlags.h: + (WTF::PtrAndFlags::PtrAndFlags): + Name lookup should not be done in dependent bases, so explicitly qualify the call to set. + +2009-12-09 Maciej Stachowiak <mjs@apple.com> + + Reviewed by Oliver Hunt. + + Google reader gets stuck in the "Loading..." state and does not complete + https://bugs.webkit.org/show_bug.cgi?id=32256 + <rdar://problem/7456388> + + * jit/JITArithmetic.cpp: + (JSC::JIT::emitSlow_op_jless): Fix some backward branches. + +2009-12-09 Gavin Barraclough <barraclough@apple.com> + + Reviewed by Oliver Hunt. + + https://bugs.webkit.org/show_bug.cgi?id=32228 + Make destruction of ropes non-recursive to prevent stack exhaustion. + Also, pass a UString& into initializeFiber rather than a Ustring::Rep*, + since the Rep is not being ref counted this could result in usage of a + Rep with refcount zero (where the Rep comes from a temporary UString + returned from a function). + + * runtime/JSString.cpp: + (JSC::JSString::Rope::destructNonRecursive): + (JSC::JSString::Rope::~Rope): + * runtime/JSString.h: + (JSC::JSString::Rope::initializeFiber): + * runtime/Operations.h: + (JSC::concatenateStrings): + +2009-12-09 Zoltan Herczeg <zherczeg@inf.u-szeged.hu> + + Reviewed by Eric Seidel. + + https://bugs.webkit.org/show_bug.cgi?id=31930 + + Update to r51457. ASSERTs changed to COMPILE_ASSERTs. + The speedup is 25%. + + * runtime/JSGlobalData.cpp: + (JSC::VPtrSet::VPtrSet): + +2009-12-09 Steve Block <steveblock@google.com> + + Reviewed by Adam Barth. + + Updates Android Makefiles with latest additions. + https://bugs.webkit.org/show_bug.cgi?id=32278 + + * Android.mk: Modified. + * Android.v8.wtf.mk: Modified. + +2009-12-09 Sam Weinig <sam@webkit.org> + + Reviewed by Gavin Barraclough. + + Fix a bug found while trying to compile JavaScriptCore with clang++. + + * yarr/RegexPattern.h: + (JSC::Yarr::PatternTerm::PatternTerm): Don't self assign here. Use false instead. + +2009-12-09 Anders Carlsson <andersca@apple.com> + + Reviewed by Sam Weinig. + + Attempt to fix the Windows build. + + * wtf/FastMalloc.h: + +2009-12-09 Anders Carlsson <andersca@apple.com> + + Reviewed by Sam Weinig. + + Fix some things found while trying to compile JavaScriptCore with clang++. + + * wtf/FastMalloc.h: + Add correct exception specifications for the allocation/deallocation operators. + + * wtf/Vector.h: + * wtf/VectorTraits.h: + Fix a bunch of struct/class mismatches. + +2009-12-08 Maciej Stachowiak <mjs@apple.com> + + Reviewed by Darin Adler. + + move code generation portions of Nodes.cpp to bytecompiler directory + https://bugs.webkit.org/show_bug.cgi?id=32284 + + * bytecompiler/NodesCodegen.cpp: Copied from parser/Nodes.cpp. Removed parts that + are not about codegen. + * parser/Nodes.cpp: Removed everything that is about codegen. + + Update build systems: + + * Android.mk: + * GNUmakefile.am: + * JavaScriptCore.gypi: + * JavaScriptCore.pri: + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.vcproj: + * JavaScriptCore.xcodeproj/project.pbxproj: + * JavaScriptCoreSources.bkl: + +2009-12-08 Kevin Watters <kevinwatters@gmail.com> + + Reviewed by Kevin Ollivier. + + [wx] Mac plugins support. + + https://bugs.webkit.org/show_bug.cgi?id=32236 + + * wtf/Platform.h: + +2009-12-08 Dmitry Titov <dimich@chromium.org> + + Rubber-stamped by David Levin. + + Revert and reopen "Add asserts to RefCounted to make sure ref/deref happens on the right thread." + It may have caused massive increase of reported leaks on the bots. + https://bugs.webkit.org/show_bug.cgi?id=31639 + + * GNUmakefile.am: + * JavaScriptCore.gypi: + * JavaScriptCore.vcproj/WTF/WTF.vcproj: + * JavaScriptCore.xcodeproj/project.pbxproj: + * runtime/Structure.cpp: + (JSC::Structure::Structure): + * wtf/RefCounted.h: + (WTF::RefCountedBase::ref): + (WTF::RefCountedBase::hasOneRef): + (WTF::RefCountedBase::refCount): + (WTF::RefCountedBase::derefBase): + * wtf/ThreadVerifier.h: Removed. + +2009-12-08 Gustavo Noronha Silva <gustavo.noronha@collabora.co.uk> + + Reviewed by Darin Adler. + + Make WebKit build correctly on FreeBSD, IA64, and Alpha. + Based on work by Petr Salinger <Petr.Salinger@seznam.cz>, + and Colin Watson <cjwatson@ubuntu.com>. + + * wtf/Platform.h: + +2009-12-08 Dmitry Titov <dimich@chromium.org> + + Reviewed by Darin Adler. + + Add asserts to RefCounted to make sure ref/deref happens on the right thread. + https://bugs.webkit.org/show_bug.cgi?id=31639 + + * runtime/Structure.cpp: + (JSC::Structure::Structure): Disable thread verification on this class since it uses addressOfCount(). + * wtf/RefCounted.h: + (WTF::RefCountedBase::ref): Add ASSERT. + (WTF::RefCountedBase::hasOneRef): Ditto. + (WTF::RefCountedBase::refCount): Ditto. + (WTF::RefCountedBase::derefBase): Ditto. + (WTF::RefCountedBase::disableThreadVerification): delegate to ThreadVerifier method. + * wtf/ThreadVerifier.h: Added. + (WTF::ThreadVerifier::ThreadVerifier): New Debug-only class to verify that ref/deref of RefCounted is done on the same thread. + (WTF::ThreadVerifier::activate): Activates checks. Called when ref count becomes above 2. + (WTF::ThreadVerifier::deactivate): Deactivates checks. Called when ref count drops below 2. + (WTF::ThreadVerifier::disableThreadVerification): used on objects that should not be checked (StringImpl etc) + (WTF::ThreadVerifier::verifyThread): + * GNUmakefile.am: Add ThreadVerifier.h to the build file. + * JavaScriptCore.gypi: Ditto. + * JavaScriptCore.vcproj/WTF/WTF.vcproj: Ditto. + * JavaScriptCore.xcodeproj/project.pbxproj: Ditto. + +2009-12-08 Steve Block <steveblock@google.com> + + Reviewed by Adam Barth. + + [Android] Adds Makefiles for Android port. + https://bugs.webkit.org/show_bug.cgi?id=31325 + + * Android.mk: Added. + * Android.v8.wtf.mk: Added. + +2009-12-07 Dmitry Titov <dimich@chromium.org> + + Rubber-stamped by Darin Adler. + + Remove ENABLE_SHARED_SCRIPT flags + https://bugs.webkit.org/show_bug.cgi?id=32245 + This patch was obtained by "git revert" command and then un-reverting of ChangeLog files. + + * Configurations/FeatureDefines.xcconfig: + * wtf/Platform.h: + +2009-12-07 Gavin Barraclough <barraclough@apple.com> + + Reviewed by NOBODY (Windows build fixage part I). + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + +2009-12-05 Gavin Barraclough <barraclough@apple.com> + + Reviewed by Oliver Hunt. + + https://bugs.webkit.org/show_bug.cgi?id=32184 + Handle out-of-memory conditions with JSC Ropes with a JS exception, rather than crashing. + Switch from using fastMalloc to tryFastMalloc, pass an ExecState to record the exception on. + + * API/JSCallbackObjectFunctions.h: + (JSC::::toString): + * API/JSValueRef.cpp: + (JSValueIsStrictEqual): + * JavaScriptCore.exp: + * bytecompiler/BytecodeGenerator.cpp: + (JSC::BytecodeGenerator::emitEqualityOp): + * debugger/DebuggerCallFrame.cpp: + (JSC::DebuggerCallFrame::functionName): + (JSC::DebuggerCallFrame::calculatedFunctionName): + * interpreter/Interpreter.cpp: + (JSC::Interpreter::callEval): + (JSC::Interpreter::privateExecute): + * jit/JITStubs.cpp: + (JSC::DEFINE_STUB_FUNCTION): + * profiler/ProfileGenerator.cpp: + (JSC::ProfileGenerator::addParentForConsoleStart): + * profiler/Profiler.cpp: + (JSC::Profiler::willExecute): + (JSC::Profiler::didExecute): + (JSC::Profiler::createCallIdentifier): + (JSC::createCallIdentifierFromFunctionImp): + * profiler/Profiler.h: + * runtime/ArrayPrototype.cpp: + (JSC::arrayProtoFuncIndexOf): + (JSC::arrayProtoFuncLastIndexOf): + * runtime/DateConstructor.cpp: + (JSC::constructDate): + * runtime/FunctionPrototype.cpp: + (JSC::functionProtoFuncToString): + * runtime/InternalFunction.cpp: + (JSC::InternalFunction::name): + (JSC::InternalFunction::displayName): + (JSC::InternalFunction::calculatedDisplayName): + * runtime/InternalFunction.h: + * runtime/JSCell.cpp: + (JSC::JSCell::getString): + * runtime/JSCell.h: + (JSC::JSValue::getString): + * runtime/JSONObject.cpp: + (JSC::gap): + (JSC::Stringifier::Stringifier): + (JSC::Stringifier::appendStringifiedValue): + * runtime/JSObject.cpp: + (JSC::JSObject::putDirectFunction): + (JSC::JSObject::putDirectFunctionWithoutTransition): + (JSC::JSObject::defineOwnProperty): + * runtime/JSObject.h: + * runtime/JSPropertyNameIterator.cpp: + (JSC::JSPropertyNameIterator::get): + * runtime/JSString.cpp: + (JSC::JSString::Rope::~Rope): + (JSC::JSString::resolveRope): + (JSC::JSString::getPrimitiveNumber): + (JSC::JSString::toNumber): + (JSC::JSString::toString): + (JSC::JSString::toThisString): + (JSC::JSString::getStringPropertyDescriptor): + * runtime/JSString.h: + (JSC::JSString::Rope::createOrNull): + (JSC::JSString::Rope::operator new): + (JSC::JSString::value): + (JSC::JSString::tryGetValue): + (JSC::JSString::getIndex): + (JSC::JSString::getStringPropertySlot): + (JSC::JSValue::toString): + * runtime/JSValue.h: + * runtime/NativeErrorConstructor.cpp: + (JSC::NativeErrorConstructor::NativeErrorConstructor): + * runtime/Operations.cpp: + (JSC::JSValue::strictEqualSlowCase): + * runtime/Operations.h: + (JSC::JSValue::equalSlowCaseInline): + (JSC::JSValue::strictEqualSlowCaseInline): + (JSC::JSValue::strictEqual): + (JSC::jsLess): + (JSC::jsLessEq): + (JSC::jsAdd): + (JSC::concatenateStrings): + * runtime/PropertyDescriptor.cpp: + (JSC::PropertyDescriptor::equalTo): + * runtime/PropertyDescriptor.h: + * runtime/StringPrototype.cpp: + (JSC::stringProtoFuncReplace): + (JSC::stringProtoFuncToLowerCase): + (JSC::stringProtoFuncToUpperCase): + +2009-12-07 Nikolas Zimmermann <nzimmermann@rim.com> + + Reviewed by Holger Freyther. + + Turn on (SVG) Filters support, by default. + https://bugs.webkit.org/show_bug.cgi?id=32224 + + * Configurations/FeatureDefines.xcconfig: Enable FILTERS build flag. + +2009-12-07 Steve Falkenburg <sfalken@apple.com> + + Build fix. Be flexible about which version of ICU is used on Windows. + + * JavaScriptCore.vcproj/jsc/jscCommon.vsprops: Add optional xcopy commands to copy ICU 4.2. + +2009-12-07 Maciej Stachowiak <mjs@apple.com> + + Reviewed by Oliver Hunt. + + op_loop_if_less JIT codegen is broken for 64-bit + https://bugs.webkit.org/show_bug.cgi?id=32221 + + * jit/JITOpcodes.cpp: + (JSC::JIT::emit_op_loop_if_false): Fix codegen in this version - test was backwards. + +2009-12-07 Oliver Hunt <oliver@apple.com> + + Reviewed by Maciej Stachowiak. + + Object.create fails if properties on the descriptor are getters + https://bugs.webkit.org/show_bug.cgi?id=32219 + + Correctly initialise the PropertySlots with the descriptor object. + + * runtime/ObjectConstructor.cpp: + (JSC::toPropertyDescriptor): + +2009-12-06 Maciej Stachowiak <mjs@apple.com> + + Not reviewed, build fix. + + Actually tested 64-bit *and* 32-bit build this time. + + * jit/JITOpcodes.cpp: + (JSC::JIT::emit_op_loop_if_false): + +2009-12-06 Maciej Stachowiak <mjs@apple.com> + + Not reviewed, build fix. + + Really really fix 64-bit build for prior patch (actually tested this time). + + * jit/JITOpcodes.cpp: + (JSC::JIT::emit_op_loop_if_false): + (JSC::JIT::emitSlow_op_loop_if_false): + +2009-12-06 Maciej Stachowiak <mjs@apple.com> + + Not reviewed, build fix. + + Really fix 64-bit build for prior patch. + + * jit/JITArithmetic.cpp: + (JSC::JIT::emitSlow_op_jless): + +2009-12-06 Maciej Stachowiak <mjs@apple.com> + + Not reviewed, build fix. + + Fix 64-bit build for prior patch. + + * jit/JITOpcodes.cpp: + (JSC::JIT::emitSlow_op_loop_if_less): + +2009-12-05 Maciej Stachowiak <mjs@apple.com> + + Reviewed by Oliver Hunt. + + conway benchmark spends half it's time in op_less (jump fusion fails) + https://bugs.webkit.org/show_bug.cgi?id=32190 + + <1% speedup on SunSpider and V8 + 2x speedup on "conway" benchmark + + Two optimizations: + 1) Improve codegen for logical operators &&, || and ! in a condition context + + When generating code for combinations of &&, || and !, in a + condition context (i.e. in an if statement or loop condition), we + used to produce a value, and then separately jump based on its + truthiness. Now we pass the false and true targets in, and let the + logical operators generate jumps directly. This helps in four + ways: + + a) Individual clauses of a short-circuit logical operator can now + jump directly to the then or else clause of an if statement (or to + the top or exit of a loop) instead of jumping to a jump. + + b) It used to be that jump fusion with the condition of the first + clause of a logical operator was inhibited, because the register + was ref'd to be used later, in the actual condition jump; this no + longer happens since a jump straight to the final target is + generated directly. + + c) It used to be that jump fusion with the condition of the second + clause of a logical operator was inhibited, because there was a + jump target right after the second clause and before the actual + condition jump. But now it's no longer necessary for the first + clause to jump there so jump fusion is not blocked. + + d) We avoid generating excess mov statements in some cases. + + As a concrete example this source: + + if (!((x < q && y < q) || (t < q && z < q))) { + // ... + } + + Used to generate this bytecode: + + [ 34] less r1, r-15, r-19 + [ 38] jfalse r1, 7(->45) + [ 41] less r1, r-16, r-19 + [ 45] jtrue r1, 14(->59) + [ 48] less r1, r-17, r-19 + [ 52] jfalse r1, 7(->59) + [ 55] less r1, r-18, r-19 + [ 59] jtrue r1, 17(->76) + + And now generates this bytecode (also taking advantage of the second optimization below): + + [ 34] jnless r-15, r-19, 8(->42) + [ 38] jless r-16, r-19, 26(->64) + [ 42] jnless r-17, r-19, 8(->50) + [ 46] jless r-18, r-19, 18(->64) + + Note the jump fusion and the fact that there's less jump + indirection - three of the four jumps go straight to the target + clause instead of indirecting through another jump. + + 2) Implement jless opcode to take advantage of the above, since we'll now often generate + a less followed by a jtrue where fusion is not forbidden. + + * parser/Nodes.h: + (JSC::ExpressionNode::hasConditionContextCodegen): Helper function to determine + whether a node supports special conditional codegen. Return false as this is the default. + (JSC::ExpressionNode::emitBytecodeInConditionContext): Assert not reached - only really + defined for nodes that do have conditional codegen. + (JSC::UnaryOpNode::expr): Add const version. + (JSC::LogicalNotNode::hasConditionContextCodegen): Returne true only if subexpression + supports it. + (JSC::LogicalOpNode::hasConditionContextCodegen): Return true. + * parser/Nodes.cpp: + (JSC::LogicalNotNode::emitBytecodeInConditionContext): Implemented - just swap + the true and false targets for the child node. + (JSC::LogicalOpNode::emitBytecodeInConditionContext): Implemented - handle jumps + directly, improving codegen quality. Also handles further nested conditional codegen. + (JSC::ConditionalNode::emitBytecode): Use condition context codegen when available. + (JSC::IfNode::emitBytecode): ditto + (JSC::IfElseNode::emitBytecode): ditto + (JSC::DoWhileNode::emitBytecode): ditto + (JSC::WhileNode::emitBytecode): ditto + (JSC::ForNode::emitBytecode): ditto + + * bytecode/Opcode.h: + - Added loop_if_false opcode - needed now that falsey jumps can be backwards. + - Added jless opcode to take advantage of new fusion opportunities. + * bytecode/CodeBlock.cpp: + (JSC::CodeBlock::dump): Handle above. + * bytecompiler/BytecodeGenerator.cpp: + (JSC::BytecodeGenerator::emitJumpIfTrue): Add peephole for less + jtrue ==> jless. + (JSC::BytecodeGenerator::emitJumpIfFalse): Add handling of backwrds falsey jumps. + * bytecompiler/BytecodeGenerator.h: + (JSC::BytecodeGenerator::emitNodeInConditionContext): Wrapper to handle tracking of + overly deep expressions etc. + * interpreter/Interpreter.cpp: + (JSC::Interpreter::privateExecute): Implement the two new opcodes (loop_if_false, jless). + * jit/JIT.cpp: + (JSC::JIT::privateCompileMainPass): Implement JIT support for the two new opcodes. + (JSC::JIT::privateCompileSlowCases): ditto + * jit/JIT.h: + * jit/JITArithmetic.cpp: + (JSC::JIT::emit_op_jless): + (JSC::JIT::emitSlow_op_jless): ditto + (JSC::JIT::emitBinaryDoubleOp): ditto + * jit/JITOpcodes.cpp: + (JSC::JIT::emitSlow_op_loop_if_less): ditto + (JSC::JIT::emit_op_loop_if_false): ditto + (JSC::JIT::emitSlow_op_loop_if_false): ditto + * jit/JITStubs.cpp: + * jit/JITStubs.h: + (JSC::): + +2009-12-04 Kent Hansen <kent.hansen@nokia.com> + + Reviewed by Darin Adler. + + JavaScript delete operator should return false for string properties + https://bugs.webkit.org/show_bug.cgi?id=32012 + + * runtime/StringObject.cpp: + (JSC::StringObject::deleteProperty): + +2009-12-03 Drew Wilson <atwilson@chromium.org> + + Rolled back r51633 because it causes a perf regression in Chromium. + + * wtf/Platform.h: + +2009-12-03 Gavin Barraclough <barraclough@apple.com> + + Try and fix the Windows build. + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: Export a symbol that should be exported. + +2009-12-03 Mark Rowe <mrowe@apple.com> + + Try and fix the Mac build. + + * JavaScriptCore.exp: Export a symbol that should be exported. + +2009-12-03 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + REGRESSION(4.0.3-48777): Crash in JSC::ExecState::propertyNames() (Debug-only?) + https://bugs.webkit.org/show_bug.cgi?id=32133 + + Work around odd GCC-ism and correct the scopechain for use by + calls made while a cachedcall is active on the callstack. + + * interpreter/CachedCall.h: + (JSC::CachedCall::newCallFrame): + * runtime/JSArray.cpp: + (JSC::AVLTreeAbstractorForArrayCompare::compare_key_key): + * runtime/StringPrototype.cpp: + (JSC::stringProtoFuncReplace): + +2009-12-03 Gavin Barraclough <barraclough@apple.com> + + Reviewed by Oliver "Brraaaaiiiinnnnnzzzzzzzz" Hunt. + + https://bugs.webkit.org/show_bug.cgi?id=32136 + Add a rope representation to JSString. Presently JSString always holds its data in UString form. + Instead, allow the result of a string concatenation to be represented in a tree form - with a + variable sized, reference-counted rope node retaining a set of UString::Reps (or other rope nopes). + + Strings must still currently be resolved down to a flat UString representation before being used, + but by holding the string in a rope representation during construction we can avoid copying data + until we know the final size of the string. + + ~2% progression on SunSpider (~25% on date-format-xparb, ~20% on string-validate-input). + + * JavaScriptCore.exp: + + - Update exports. + + * interpreter/Interpreter.cpp: + (JSC::Interpreter::privateExecute): + + - Make use of new JSString::length() method to avoid prematurely resolving ropes. + + * jit/JITOpcodes.cpp: + (JSC::JIT::privateCompileCTIMachineTrampolines): + + - Switch the string length trampoline to read the length directly from JSString::m_length, + rather than from the JSString's UString::Rep's 'len' property. + + * jit/JITStubs.cpp: + (JSC::DEFINE_STUB_FUNCTION): + + - Modify op_add such that addition of two strings, where either or both strings are already + in rope representation, produces a rope as a result. + + * runtime/JSString.cpp: + (JSC::JSString::Rope::~Rope): + (JSC::copyChars): + (JSC::JSString::resolveRope): + (JSC::JSString::getPrimitiveNumber): + (JSC::JSString::toBoolean): + (JSC::JSString::toNumber): + (JSC::JSString::toString): + (JSC::JSString::toThisString): + (JSC::JSString::getStringPropertyDescriptor): + * runtime/JSString.h: + (JSC::JSString::Rope::Fiber::Fiber): + (JSC::JSString::Rope::Fiber::destroy): + (JSC::JSString::Rope::Fiber::isRope): + (JSC::JSString::Rope::Fiber::rope): + (JSC::JSString::Rope::Fiber::string): + (JSC::JSString::Rope::create): + (JSC::JSString::Rope::initializeFiber): + (JSC::JSString::Rope::ropeLength): + (JSC::JSString::Rope::stringLength): + (JSC::JSString::Rope::fibers): + (JSC::JSString::Rope::Rope): + (JSC::JSString::Rope::operator new): + (JSC::JSString::JSString): + (JSC::JSString::value): + (JSC::JSString::length): + (JSC::JSString::isRope): + (JSC::JSString::rope): + (JSC::JSString::string): + (JSC::JSString::canGetIndex): + (JSC::jsSingleCharacterSubstring): + (JSC::JSString::getIndex): + (JSC::jsSubstring): + (JSC::JSString::getStringPropertySlot): + + - Add rope form. + + * runtime/Operations.h: + (JSC::jsAdd): + (JSC::concatenateStrings): + + - Update string concatenation, and addition of ropes, to produce ropes. + + * runtime/StringObject.cpp: + (JSC::StringObject::getOwnPropertyNames): + + - Make use of new JSString::length() method to avoid prematurely resolving ropes. + +2009-11-23 Jeremy Moskovich <jeremy@chromium.org> + + Reviewed by Eric Seidel. + + Switch Chrome/Mac to use Core Text APIs rather than ATSUI APIs. + https://bugs.webkit.org/show_bug.cgi?id=31802 + + No test since this is already covered by existing pixel tests. + + * wtf/Platform.h: #define USE_CORE_TEXT for Chrome/Mac. + +2009-12-02 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + Add files missed in prior patch. + + * runtime/JSZombie.cpp: + (JSC::): + (JSC::JSZombie::leakedZombieStructure): + * runtime/JSZombie.h: Added. + (JSC::JSZombie::JSZombie): + (JSC::JSZombie::isZombie): + (JSC::JSZombie::classInfo): + (JSC::JSZombie::isGetterSetter): + (JSC::JSZombie::isAPIValueWrapper): + (JSC::JSZombie::isPropertyNameIterator): + (JSC::JSZombie::getCallData): + (JSC::JSZombie::getConstructData): + (JSC::JSZombie::getUInt32): + (JSC::JSZombie::toPrimitive): + (JSC::JSZombie::getPrimitiveNumber): + (JSC::JSZombie::toBoolean): + (JSC::JSZombie::toNumber): + (JSC::JSZombie::toString): + (JSC::JSZombie::toObject): + (JSC::JSZombie::markChildren): + (JSC::JSZombie::put): + (JSC::JSZombie::deleteProperty): + (JSC::JSZombie::toThisObject): + (JSC::JSZombie::toThisString): + (JSC::JSZombie::toThisJSString): + (JSC::JSZombie::getJSNumber): + (JSC::JSZombie::getOwnPropertySlot): + +2009-12-02 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + Add zombies to JSC + https://bugs.webkit.org/show_bug.cgi?id=32103 + + Add a compile time flag to make the JSC collector replace "unreachable" + objects with zombie objects. The zombie object is a JSCell subclass that + ASSERTs on any attempt to use the JSCell methods. In addition there are + a number of additional assertions in bottleneck code to catch zombie usage + as quickly as possible. + + Grrr. Argh. Brains. + + * JavaScriptCore.xcodeproj/project.pbxproj: + * interpreter/Register.h: + (JSC::Register::Register): + * runtime/ArgList.h: + (JSC::MarkedArgumentBuffer::append): + (JSC::ArgList::ArgList): + * runtime/Collector.cpp: + (JSC::Heap::destroy): + (JSC::Heap::sweep): + * runtime/Collector.h: + * runtime/JSCell.h: + (JSC::JSCell::isZombie): + (JSC::JSValue::isZombie): + * runtime/JSValue.h: + (JSC::JSValue::decode): + (JSC::JSValue::JSValue): + * wtf/Platform.h: + +2009-12-01 Jens Alfke <snej@chromium.org> + + Reviewed by Darin Adler. + + Added variants of find/contains/add that allow a foreign key type to be used. + This will allow AtomicString-keyed maps to be queried by C string without + having to create a temporary AtomicString (see HTTPHeaderMap.) + The code for this is adapted from the equivalent in HashSet.h. + + * wtf/HashMap.h: + (WTF::HashMap::find): + (WTF::HashMap::contains): + (WTF::HashMap::add): + * wtf/HashSet.h: Changed "method" to "function member" in a comment. + +2009-12-01 Gustavo Noronha Silva <gustavo.noronha@collabora.co.uk> + + Revert 51551 because it broke GTK+. + + * wtf/Platform.h: + +2009-11-30 Gavin Barraclough <barraclough@apple.com> + + Windows Build fix. Reviewed by NOBODY. + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + +2009-11-24 Gavin Barraclough <barraclough@apple.com> + + Reviewed by Geoff Garen. + + Bug 31859 - Make world selection for JSC IsolatedWorlds automagical. + + WebCore presently has to explicitly specify the world before entering into JSC, + which is a little fragile (particularly since property access via a + getter/setter might invoke execution). Instead derive the current world from + the lexical global object. + + Remove the temporary duct tape of willExecute/didExecute virtual hooks on the JSGlobalData::ClientData - these are no longer necessary. + + * API/JSBase.cpp: + (JSEvaluateScript): + * API/JSObjectRef.cpp: + (JSObjectCallAsFunction): + * JavaScriptCore.exp: + * runtime/JSGlobalData.cpp: + * runtime/JSGlobalData.h: + +2009-11-30 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Kenneth Rohde Christiansen. + + [Qt] Remove obsolete PLATFORM(KDE) code + https://bugs.webkit.org/show_bug.cgi?id=31958 + + KDE is now using unpatched QtWebKit. + + * parser/Lexer.cpp: Remove obsolete KDE_USE_FINAL guard + * wtf/Platform.h: Remove PLATFORM(KDE) definition and code + section that is guarded with it. + +2009-11-30 Jan-Arve Sæther <jan-arve.saether@nokia.com> + + Reviewed by Simon Hausmann. + + [Qt] Fix compilation with win32-icc + + The Intel compiler does not support the __has_trivial_constructor type + trait. The Intel Compiler can report itself as _MSC_VER >= 1400. The + reason for that is that the Intel Compiler depends on the Microsoft + Platform SDK, and in order to try to be "fully" MS compatible it will + "pretend" to be the same MS compiler as was shipped with the MS PSDK. + (Thus, compiling with win32-icc with VC8 SDK will make the source code + "think" the compiler at hand supports this type trait). + + * wtf/TypeTraits.h: + +2009-11-29 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Eric Seidel. + + [Qt] Mac build has JIT disabled + https://bugs.webkit.org/show_bug.cgi?id=31828 + + * wtf/Platform.h: Enable JIT for Qt Mac builds + +2009-11-28 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Eric Seidel. + + Apply workaround for the limitation of VirtualFree with MEM_RELEASE to all ports running on Windows + https://bugs.webkit.org/show_bug.cgi?id=31943 + + * runtime/MarkStack.h: + (JSC::MarkStack::MarkStackArray::shrinkAllocation): + +2009-11-28 Zoltan Herczeg <zherczeg@inf.u-szeged.hu> + + Reviewed by Gavin Barraclough. + + https://bugs.webkit.org/show_bug.cgi?id=31930 + + Seems a typo. We don't need ~270k memory to determine the vptrs. + + * runtime/JSGlobalData.cpp: + (JSC::VPtrSet::VPtrSet): + +2009-11-27 Shinichiro Hamaji <hamaji@chromium.org> + + Unreviewed. + + Move GOwnPtr* from wtf to wtf/gtk + https://bugs.webkit.org/show_bug.cgi?id=31793 + + Build fix for chromium after r51423. + Exclude gtk directory from chromium build. + + * JavaScriptCore.gyp/JavaScriptCore.gyp: + +2009-11-25 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + Incorrect behaviour of jneq_null in the interpreter + https://bugs.webkit.org/show_bug.cgi?id=31901 + + Correct the logic of jneq_null. This is already covered by existing tests. + + * interpreter/Interpreter.cpp: + (JSC::Interpreter::privateExecute): + +2009-11-26 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Oliver Hunt. + + Move GOwnPtr* from wtf to wtf/gtk + https://bugs.webkit.org/show_bug.cgi?id=31793 + + * GNUmakefile.am: Change the path for GOwnPtr.*. + * JavaScriptCore.gyp/JavaScriptCore.gyp: Remove + GOwnPtr.cpp from the exclude list. + * JavaScriptCore.gypi: Change the path for GOwnPtr.*. + * wscript: Remove GOwnPtr.cpp from the exclude list. + * wtf/GOwnPtr.cpp: Removed. + * wtf/GOwnPtr.h: Removed. + * wtf/Threading.h: Change the path for GOwnPtr.h. + * wtf/gtk/GOwnPtr.cpp: Copied from JavaScriptCore/wtf/GOwnPtr.cpp. + * wtf/gtk/GOwnPtr.h: Copied from JavaScriptCore/wtf/GOwnPtr.h. + * wtf/unicode/glib/UnicodeGLib.h: Change the path for GOwnPtr.h. + +2009-11-24 Dmitry Titov <dimich@chromium.org> + + Reviewed by Eric Seidel. + + Add ENABLE_SHARED_SCRIPT feature define and flag for build-webkit + https://bugs.webkit.org/show_bug.cgi?id=31444 + + * Configurations/FeatureDefines.xcconfig: + * wtf/Platform.h: + +2009-11-24 Chris Marrin <cmarrin@apple.com> + + Reviewed by Simon Fraser. + + Add ability to enable ACCELERATED_COMPOSITING on Windows (currently disabled) + https://bugs.webkit.org/show_bug.cgi?id=27314 + + * wtf/Platform.h: + +2009-11-24 Jason Smith <dark.panda@gmail.com> + + Reviewed by Alexey Proskuryakov. + + RegExp#exec's returned Array-like object behaves differently from + regular Arrays + https://bugs.webkit.org/show_bug.cgi?id=31689 + + * JavaScriptCore/runtime/RegExpConstructor.cpp: ensure that undefined + values are added to the returned RegExpMatchesArray + +2009-11-24 Oliver Hunt <oliver@apple.com> + + Reviewed by Alexey Proskuryakov. + + JSON.stringify performance on undefined is very poor + https://bugs.webkit.org/show_bug.cgi?id=31839 + + Switch from a UString to a Vector<UChar> when building + the JSON string, allowing us to safely remove the substr-copy + we otherwise did when unwinding an undefined property. + + Also turns out to be a ~5% speedup on stringification. + + * runtime/JSONObject.cpp: + (JSC::Stringifier::StringBuilder::append): + (JSC::Stringifier::stringify): + (JSC::Stringifier::Holder::appendNextProperty): + +2009-11-24 Mark Rowe <mrowe@apple.com> + + Fix production builds where the source tree may be read-only. + + * JavaScriptCore.xcodeproj/project.pbxproj: + +2009-11-23 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Kenneth Rohde Christiansen. + + Include "config.h" to meet Coding Style Guidelines + https://bugs.webkit.org/show_bug.cgi?id=31792 + + * wtf/unicode/UTF8.cpp: + * wtf/unicode/glib/UnicodeGLib.cpp: + * wtf/unicode/wince/UnicodeWince.cpp: + +2009-11-23 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Oliver Hunt. + + Streamlined some Math functions where we expect or know the result not + to be representable as an int. + + SunSpider says 0.6% faster. + + * runtime/JSNumberCell.h: + (JSC::JSValue::JSValue): + * runtime/JSValue.h: + (JSC::JSValue::): + (JSC::jsDoubleNumber): + (JSC::JSValue::JSValue): Added a function for making a numeric JSValue + and skipping the "can I encode this as an int?" check, avoiding the + overhead of int <-> double roundtripping and double <-> double comparison + and branching. + + * runtime/MathObject.cpp: + (JSC::mathProtoFuncACos): + (JSC::mathProtoFuncASin): + (JSC::mathProtoFuncATan): + (JSC::mathProtoFuncATan2): + (JSC::mathProtoFuncCos): + (JSC::mathProtoFuncExp): + (JSC::mathProtoFuncLog): + (JSC::mathProtoFuncRandom): + (JSC::mathProtoFuncSin): + (JSC::mathProtoFuncSqrt): + (JSC::mathProtoFuncTan): For these functions, which we expect or know + to produce results not representable as ints, call jsDoubleNumber instead + of jsNumber. + +2009-11-23 Mark Rowe <mrowe@apple.com> + + Unreviewed. Unbreak the regression tests after r51329. + + * API/JSBase.cpp: + (JSEvaluateScript): Null-check clientData before dereferencing it. + * API/JSObjectRef.cpp: + (JSObjectCallAsFunction): Ditto. + +2009-11-23 Gavin Barraclough <barraclough@apple.com> + + Reviewed by Geoff Garen. + + Part 1/3 of <rdar://problem/7377477> REGRESSION: Many web pages fail to render after interesting script runs in isolated world + + Some clients of the JavaScriptCore API expect to be able to make callbacks over the JSC API, + and for this to automagically cause execution to take place in the world associated with the + global object associated with the ExecState (JSContextRef) passed. However this is not how + things work - the world must be explicitly set within WebCore. + + Making this work just for API calls to evaluate & call will be a far from perfect solution, + since direct (non-API) use of JSC still relies on WebCore setting the current world correctly. + A better solution would be to make this all work automagically all throughout WebCore, but this + will require more refactoring. + + Since the API is in JSC but worlds only exist in WebCore, add callbacks on the JSGlobalData::ClientData + to allow it to update the current world on entry/exit via the JSC API. This is temporary duck + tape, and should be removed once the current world no longer needs to be explicitly tracked. + + * API/JSBase.cpp: + (JSEvaluateScript): + * API/JSObjectRef.cpp: + (JSObjectCallAsFunction): + * JavaScriptCore.exp: + * runtime/JSGlobalData.cpp: + (JSC::JSGlobalData::ClientData::beginningExecution): + (JSC::JSGlobalData::ClientData::completedExecution): + * runtime/JSGlobalData.h: + +2009-11-23 Steve Block <steveblock@google.com> + + Reviewed by Dmitry Titov. + + Adds MainThreadAndroid.cpp with Android-specific WTF threading functions. + https://bugs.webkit.org/show_bug.cgi?id=31807 + + * wtf/android: Added. + * wtf/android/MainThreadAndroid.cpp: Added. + (WTF::timeoutFired): + (WTF::initializeMainThreadPlatform): + (WTF::scheduleDispatchFunctionsOnMainThread): + +2009-11-23 Alexey Proskuryakov <ap@apple.com> + + Reviewed by Brady Eidson. + + https://bugs.webkit.org/show_bug.cgi?id=31748 + Make WebSocketHandleCFNet respect proxy auto-configuration files via CFProxySupport + + * JavaScriptCore.exp: Export callOnMainThreadAndWait. + +2009-11-23 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Kenneth Rohde Christiansen. + + [Symbian] Fix lastIndexOf() for Symbian + https://bugs.webkit.org/show_bug.cgi?id=31773 + + Symbian soft floating point library has problems with operators + comparing NaN to numbers. Without a workaround lastIndexOf() + function does not work. + + Patch developed by David Leong. + + * runtime/StringPrototype.cpp: + (JSC::stringProtoFuncLastIndexOf):Add an extra test + to check for NaN for Symbian. + +2009-11-23 Steve Block <steveblock@google.com> + + Reviewed by Eric Seidel. + + Android port lacks implementation of atomicIncrement and atomicDecrement. + https://bugs.webkit.org/show_bug.cgi?id=31715 + + * wtf/Threading.h: Modified. + (WTF::atomicIncrement): Added Android implementation. + (WTF::atomicDecrement): Added Android implementation. + +2009-11-22 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Unreviewed. + + [Qt] Sort source lists and remove obsolete comments + from the build system. + + * JavaScriptCore.pri: + +2009-11-21 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Eric Seidel. + + [Qt][Mac] Turn on multiple JavaScript threads for QtWebkit on Mac + https://bugs.webkit.org/show_bug.cgi?id=31753 + + * wtf/Platform.h: + +2009-11-19 Steve Block <steveblock@google.com> + + Android port lacks configuration in Platform.h and config.h. + https://bugs.webkit.org/show_bug.cgi?id=31671 + + * wtf/Platform.h: Modified. Added Android-specific configuration. + +2009-11-19 Alexey Proskuryakov <ap@apple.com> + + Reviewed by Darin Adler. + + https://bugs.webkit.org/show_bug.cgi?id=31690 + Make SocketStreamHandleCFNet work on Windows + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + * wtf/MainThread.cpp: + (WTF::FunctionWithContext::FunctionWithContext): + (WTF::dispatchFunctionsFromMainThread): + (WTF::callOnMainThreadAndWait): + * wtf/MainThread.h: + Re-add callOnMainThreadAndWait(), which was removed in bug 23926. + +2009-11-19 Dmitry Titov <dimich@chromium.org> + + Reviewed by David Levin. + + isMainThread() on Chromium (Mac and Linux) is so slow it timeouts LayoutTests.. + https://bugs.webkit.org/show_bug.cgi?id=31693 + + * wtf/ThreadingPthreads.cpp: + (WTF::initializeThreading): grab and use the pthread_t of the main thread instead of ThreadIdentifier. + (WTF::isMainThread): Ditto. + +2009-11-19 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Darin Adler. + + Remove HAVE(STRING_H) guard from JavaScriptCore + https://bugs.webkit.org/show_bug.cgi?id=31668 + + * config.h: + * runtime/UString.cpp: + +2009-11-19 Dumitru Daniliuc <dumi@chromium.org> + + Reviewed by Dmitry Titov. + + Fixing a bug in MessageQueue::removeIf() that leads to an + assertion failure. + + https://bugs.webkit.org/show_bug.cgi?id=31657 + + * wtf/MessageQueue.h: + (WTF::MessageQueue::removeIf): + +2009-11-19 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Darin Adler. + + Remove HAVE(FLOAT_H) guard + https://bugs.webkit.org/show_bug.cgi?id=31661 + + JavaScriptCore has a dependency on float.h, there is + no need to guard float.h. + + * runtime/DatePrototype.cpp: Remove include directive + for float.h as it is included in MathExtras.h already. + * runtime/Operations.cpp: Ditto. + * runtime/UString.cpp: Ditto. + * wtf/dtoa.cpp: Ditto. + * wtf/MathExtras.h: Remove HAVE(FLOAT_H) guard. + * wtf/Platform.h: Ditto. + +2009-11-19 Thiago Macieira <thiago.macieira@nokia.com> + + Reviewed by Simon Hausmann. + + Build fix for 32-bit Sparc machines: these machines are big-endian. + + * wtf/Platform.h: + +2009-11-18 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Kenneth Rohde Christiansen. + + [Qt] Remove support for Qt v4.3 or older versions + https://bugs.webkit.org/show_bug.cgi?id=29469 + + * JavaScriptCore.pro: + * jsc.pro: + * wtf/unicode/qt4/UnicodeQt4.h: + +2009-11-18 Kent Tamura <tkent@chromium.org> + + Reviewed by Darin Adler. + + Move UString::from(double) implementation to new + WTF::doubleToStringInJavaScriptFormat(), and expose it because WebCore + code will use it. + https://bugs.webkit.org/show_bug.cgi?id=31330 + + - Introduce new function createRep(const char*, unsigned) and + UString::UString(const char*, unsigned) to reduce 2 calls to strlen(). + - Fix a bug that dtoa() doesn't update *rve if the input value is NaN + or Infinity. + + No new tests because this doesn't change the behavior. + + * JavaScriptCore.exp: + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + * runtime/UString.cpp: + (JSC::createRep): + (JSC::UString::UString): + (JSC::UString::from): Move the code to doubleToStringInJavaScriptFormat(). + * runtime/UString.h: + * wtf/dtoa.cpp: + (WTF::dtoa): Fix a bug about rve. + (WTF::append): A helper for doubleToStringInJavaScriptFormat(). + (WTF::doubleToStringInJavaScriptFormat): Move the code from UString::from(double). + * wtf/dtoa.h: + +2009-11-18 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Kenneth Rohde Christiansen. + + [Qt] Remove WTF_USE_JAVASCRIPTCORE_BINDINGS as it is no longer used + https://bugs.webkit.org/show_bug.cgi?id=31643 + + * JavaScriptCore.pro: + +2009-11-18 Nate Chapin <japhet@chromium.org> + + Reviewed by Darin Fisher. + + Remove Chromium's unnecessary dependency on wtf's tcmalloc files. + + https://bugs.webkit.org/show_bug.cgi?id=31648 + + * JavaScriptCore.gyp/JavaScriptCore.gyp: + +2009-11-18 Thiago Macieira <thiago.macieira@nokia.com> + + Reviewed by Gavin Barraclough. + + [Qt] Implement symbol hiding for JSC's JIT functions. + + These functions are implemented directly in assembly, so they need the + proper directives to enable/disable visibility. On ELF systems, it's + .hidden, whereas on Mach-O systems (Mac) it's .private_extern. On + Windows, it's not necessary since you have to explicitly export. I + also implemented the AIX idiom, though it's unlikely anyone will + implement AIX/POWER JIT. + https://bugs.webkit.org/show_bug.cgi?id=30864 + + * jit/JITStubs.cpp: + +2009-11-18 Oliver Hunt <oliver@apple.com> + + Reviewed by Alexey Proskuryakov. + + Interpreter may do an out of range access when throwing an exception in the profiler. + https://bugs.webkit.org/show_bug.cgi?id=31635 + + Add bounds check. + + * interpreter/Interpreter.cpp: + (JSC::Interpreter::throwException): + +2009-11-18 Gabor Loki <loki@inf.u-szeged.hu> + + Reviewed by Darin Adler. + + Fix the clobber list of cacheFlush for ARM and Thumb2 on Linux + https://bugs.webkit.org/show_bug.cgi?id=31631 + + * jit/ExecutableAllocator.h: + (JSC::ExecutableAllocator::cacheFlush): + +2009-11-18 Harald Fernengel <harald.fernengel@nokia.com> + + Reviewed by Simon Hausmann. + + [Qt] Fix detection of linux-g++ + + Never use "linux-g++*" to check for linux-g++, since this will break embedded + builds which use linux-arm-g++ and friends. Use 'linux*-g++*' to check for any + g++ on linux mkspec. + + * JavaScriptCore.pri: + +2009-11-17 Jon Honeycutt <jhoneycutt@apple.com> + + Add JSContextRefPrivate.h to list of copied files. + + Reviewed by Mark Rowe. + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGenerated.make: + +2009-11-17 Martin Robinson <martin.james.robinson@gmail.com> + + Reviewed by Adam Barth. + + [GTK] Style cleanup for GOwnPtr + https://bugs.webkit.org/show_bug.cgi?id=31506 + + Remove forward declaration in GOwnPtr and do some style cleanup. + + * wtf/GOwnPtr.cpp: + * wtf/GOwnPtr.h: + (WTF::GOwnPtr::GOwnPtr): + (WTF::GOwnPtr::~GOwnPtr): + (WTF::GOwnPtr::get): + (WTF::GOwnPtr::release): + (WTF::GOwnPtr::outPtr): + (WTF::GOwnPtr::set): + (WTF::GOwnPtr::clear): + (WTF::GOwnPtr::operator*): + (WTF::GOwnPtr::operator->): + (WTF::GOwnPtr::operator!): + (WTF::GOwnPtr::operator UnspecifiedBoolType): + (WTF::GOwnPtr::swap): + (WTF::swap): + (WTF::operator==): + (WTF::operator!=): + (WTF::getPtr): + (WTF::freeOwnedGPtr): + +2009-11-17 Oliver Hunt <oliver@apple.com> + + Reviewed by Maciej Stachowiak. + + Incorrect use of JavaScriptCore API in DumpRenderTree + https://bugs.webkit.org/show_bug.cgi?id=31577 + + Add assertions to the 'toJS' functions to catch mistakes like + this early. Restructure existing code which blindly passed potentially + null values to toJS when forwarding exceptions so that a null check is + performed first. + + * API/APICast.h: + (toJS): + (toJSForGC): + * API/JSCallbackObjectFunctions.h: + (JSC::::getOwnPropertySlot): + (JSC::::put): + (JSC::::deleteProperty): + (JSC::::construct): + (JSC::::hasInstance): + (JSC::::call): + (JSC::::toNumber): + (JSC::::toString): + (JSC::::staticValueGetter): + (JSC::::callbackGetter): + * API/tests/testapi.c: Fix errors in the API tester. + (MyObject_getProperty): + (MyObject_convertToType): + (EvilExceptionObject_convertToType): + +2009-11-16 Zoltan Herczeg <zherczeg@inf.u-szeged.hu> + + Reviewed by Gavin Barraclough. + + https://bugs.webkit.org/show_bug.cgi?id=31050 + + Minor fixes for JSVALUE32_64: branchConvertDoubleToInt32 + failed on a CortexA8 CPU, but not on a simulator; and + JITCall.cpp modifications was somehow not committed to mainline. + + * assembler/ARMAssembler.h: + (JSC::ARMAssembler::fmrs_r): + * assembler/MacroAssemblerARM.h: + (JSC::MacroAssemblerARM::branchConvertDoubleToInt32): + * jit/JITCall.cpp: + (JSC::JIT::compileOpCall): + +2009-11-16 Joerg Bornemann <joerg.bornemann@trolltech.com> + + Reviewed by Simon Hausmann. + + Fix Qt build on Windows CE 6. + + * JavaScriptCore.pri: Add missing include path. + * wtf/Platform.h: Include ce_time.h for Windows CE 6. + +2009-11-13 Zoltan Herczeg <zherczeg@inf.u-szeged.hu> + + Reviewed by Gavin Barraclough. + + https://bugs.webkit.org/show_bug.cgi?id=31050 + + Adding optimization support for mode JSVALUE32_64 + on ARM systems. + + * jit/JIT.h: + * jit/JITCall.cpp: + (JSC::JIT::compileOpCall): + * jit/JITPropertyAccess.cpp: + (JSC::JIT::emit_op_method_check): + (JSC::JIT::compileGetByIdHotPath): + (JSC::JIT::compileGetByIdSlowCase): + (JSC::JIT::emit_op_put_by_id): + +2009-11-14 Zoltan Herczeg <zherczeg@inf.u-szeged.hu> + + Reviewed by Gavin Barraclough. + + https://bugs.webkit.org/show_bug.cgi?id=31050 + + Adding JSVALUE32_64 support for ARM (but not turning it + on by default). All optimizations must be disabled, since + this patch is only the first of a series of patches. + + During the work, a lot of x86 specific code revealed and + made platform independent. + See revisions: 50531 50541 50593 50594 50595 + + * assembler/ARMAssembler.h: + (JSC::ARMAssembler::): + (JSC::ARMAssembler::fdivd_r): + * assembler/MacroAssemblerARM.h: + (JSC::MacroAssemblerARM::lshift32): + (JSC::MacroAssemblerARM::neg32): + (JSC::MacroAssemblerARM::rshift32): + (JSC::MacroAssemblerARM::branchOr32): + (JSC::MacroAssemblerARM::set8): + (JSC::MacroAssemblerARM::setTest8): + (JSC::MacroAssemblerARM::loadDouble): + (JSC::MacroAssemblerARM::divDouble): + (JSC::MacroAssemblerARM::convertInt32ToDouble): + (JSC::MacroAssemblerARM::zeroDouble): + * jit/JIT.cpp: + * jit/JIT.h: + * jit/JITOpcodes.cpp: + (JSC::JIT::privateCompileCTIMachineTrampolines): + * jit/JITStubs.cpp: + * wtf/StdLibExtras.h: + +2009-11-13 Dominik Röttsches <dominik.roettsches@access-company.com> + + Reviewed by Eric Seidel. + + Unify TextBoundaries implementations by only relying on WTF Unicode abstractions + https://bugs.webkit.org/show_bug.cgi?id=31468 + + Adding isAlphanumeric abstraction, required + by TextBoundaries.cpp. + + * wtf/unicode/glib/UnicodeGLib.h: + (WTF::Unicode::isAlphanumeric): + * wtf/unicode/icu/UnicodeIcu.h: + (WTF::Unicode::isAlphanumeric): + +2009-11-13 Norbert Leser <norbert.leser&nokia.com> + + Reviewed by Eric Seidel. + + Added macros for USERINCLUDE paths within symbian blocks + to guarantee inclusion of respective header files from local path + first (to avoid clashes with same names of header files in system include path). + + * JavaScriptCore.pri: + +2009-11-13 Oliver Hunt <oliver@apple.com> + + Reviewed by Geoff Garen. + + JSValueProtect and JSValueUnprotect don't protect API wrapper values + https://bugs.webkit.org/show_bug.cgi?id=31485 + + Make JSValueProtect/Unprotect use a new 'toJS' function, 'toJSForGC' that + does not attempt to to strip out API wrapper objects. + + * API/APICast.h: + (toJSForGC): + * API/JSValueRef.cpp: + (JSValueProtect): + (JSValueUnprotect): + * API/tests/testapi.c: + (makeGlobalNumberValue): + (main): + +2009-11-13 İsmail Dönmez <ismail@namtrac.org> + + Reviewed by Antti Koivisto. + + Fix typo, ce_time.cpp should be ce_time.c + + * JavaScriptCore.pri: + +2009-11-12 Steve VanDeBogart <vandebo@chromium.org> + + Reviewed by Adam Barth. + + Calculate the time offset only if we were able to parse + the date string. This saves an IPC in Chromium for + invalid date strings. + https://bugs.webkit.org/show_bug.cgi?id=31416 + + * wtf/DateMath.cpp: + (WTF::parseDateFromNullTerminatedCharacters): + (JSC::parseDateFromNullTerminatedCharacters): + +2009-11-12 Oliver Hunt <oliver@apple.com> + + Rollout r50896 until i can work out why it causes failures. + + * bytecompiler/BytecodeGenerator.cpp: + (JSC::BytecodeGenerator::emitReturn): + * interpreter/Interpreter.cpp: + (JSC::Interpreter::execute): + * parser/Nodes.cpp: + (JSC::EvalNode::emitBytecode): + +2009-11-12 Steve Falkenburg <sfalken@apple.com> + + Reviewed by Stephanie Lewis. + + Remove LIBRARY directive from def file to fix Debug_All target. + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + +2009-11-12 Gustavo Noronha Silva <gustavo.noronha@collabora.co.uk> + + Rubber-stamped by Holger Freyther. + + Revert r50204, since it makes DRT crash on 32 bits release builds + for GTK+. + + * wtf/FastMalloc.h: + +2009-11-12 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + Start unifying entry logic for function and eval code. + + Eval now uses a ret instruction to end execution, and sets up + a callframe more in line with what we do for function entry. + + * bytecompiler/BytecodeGenerator.cpp: + (JSC::BytecodeGenerator::emitReturn): + * interpreter/Interpreter.cpp: + (JSC::Interpreter::execute): + * parser/Nodes.cpp: + (JSC::EvalNode::emitBytecode): + +2009-11-12 Richard Moe Gustavsen <richard.gustavsen@nokia.com> + + Reviewed by Kenneth Rohde Christiansen. + + [Qt] Disable pthread_setname_np. + + This allows Qt builds on Mac from 10.6 to run on earlier version + where this symbol is not present. + https://bugs.webkit.org/show_bug.cgi?id=31403 + + * wtf/Platform.h: + +2009-11-12 Thiago Macieira <thiago.macieira@nokia.com> + + Reviewed by Kenneth Rohde Christiansen. + + [Qt] Fix linking on Linux 32-bit. + + It was missing the ".text" directive at the top of the file, + indicating that code would follow. Without it, the assembler created + "NOTYPE" symbols, which would result in linker errors. + https://bugs.webkit.org/show_bug.cgi?id=30863 + + * jit/JITStubs.cpp: + +2009-11-11 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Alexey Proskuryakov. + + Refactor multiple JavaScriptCore threads + https://bugs.webkit.org/show_bug.cgi?id=31328 + + Remove the id field from the PlatformThread structure + as it is not used. + + * runtime/Collector.cpp: + (JSC::getCurrentPlatformThread): + (JSC::suspendThread): + (JSC::resumeThread): + (JSC::getPlatformThreadRegisters): + +2009-11-10 Geoffrey Garen <ggaren@apple.com> + + Linux build fix: Added an #include for UINT_MAX. + + * runtime/WeakRandom.h: + +2009-11-10 Geoffrey Garen <ggaren@apple.com> + + JavaScriptGlue build fix: Marked a file 'private' instead of 'project'. + + * JavaScriptCore.xcodeproj/project.pbxproj: + +2009-11-10 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Gavin "avGni arBalroguch" Barraclough. + + Faster Math.random, based on GameRand. + + SunSpider says 1.4% faster. + + * GNUmakefile.am: + * JavaScriptCore.gypi: + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.vcproj: + * JavaScriptCore.xcodeproj/project.pbxproj: Added the header to the project. + + * runtime/JSGlobalData.cpp: + (JSC::JSGlobalData::JSGlobalData): + * runtime/JSGlobalData.h: Use an object to track random number generation + state, initialized to the current time. + + * runtime/MathObject.cpp: + (JSC::MathObject::MathObject): + (JSC::mathProtoFuncRandom): Use the new hotness. + + * runtime/WeakRandom.h: Added. + (JSC::WeakRandom::WeakRandom): + (JSC::WeakRandom::get): + (JSC::WeakRandom::advance): The new hotness. + +2009-11-09 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Oliver Hunt. + + Imported the v8 DST cache. + + SunSpider says 1.5% faster. + + * runtime/JSGlobalData.cpp: + (JSC::JSGlobalData::resetDateCache): Reset the DST cache when resetting + other date data. + + * runtime/JSGlobalData.h: + (JSC::DSTOffsetCache::DSTOffsetCache): + (JSC::DSTOffsetCache::reset): Added a struct for the DST cache. + + * wtf/DateMath.cpp: + (WTF::calculateDSTOffsetSimple): + (WTF::calculateDSTOffset): + (WTF::parseDateFromNullTerminatedCharacters): + (JSC::getDSTOffset): + (JSC::gregorianDateTimeToMS): + (JSC::msToGregorianDateTime): + (JSC::parseDateFromNullTerminatedCharacters): + * wtf/DateMath.h: The imported code for probing and updating the cache. + +2009-11-09 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Oliver Hunt. + + Fixed an edge case that could cause the engine not to notice a timezone + change. + + No test because this case would require manual intervention to change + the timezone during the test. + + SunSpider reports no change. + + * runtime/DateInstanceCache.h: + (JSC::DateInstanceCache::DateInstanceCache): + (JSC::DateInstanceCache::reset): Added a helper function for resetting + this cache. Also, shrank the cache, since we'll be resetting it often. + + * runtime/JSGlobalData.cpp: + (JSC::JSGlobalData::resetDateCache): Include resetting the DateInstanceCache + in resetting Date data. (Otherwise, a cache hit could bypass a necessary + timezone update check.) + +2009-11-09 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Sam Weinig. + + Some manual inlining and constant propogation in Date code. + + SunSpider reports a 0.4% speedup on date-*, no overall speedup. Shark + says some previously evident stalls are now gone. + + * runtime/DateConstructor.cpp: + (JSC::callDate): + * runtime/DateConversion.cpp: + (JSC::formatTime): + (JSC::formatTimeUTC): Split formatTime into UTC and non-UTC variants. + + * runtime/DateConversion.h: + * runtime/DateInstance.cpp: + (JSC::DateInstance::calculateGregorianDateTime): + (JSC::DateInstance::calculateGregorianDateTimeUTC): + * runtime/DateInstance.h: + (JSC::DateInstance::gregorianDateTime): + (JSC::DateInstance::gregorianDateTimeUTC): Split gregorianDateTime into + a UTC and non-UTC variant, and split each variant into a fast inline + case and a slow out-of-line case. + + * runtime/DatePrototype.cpp: + (JSC::formatLocaleDate): + (JSC::dateProtoFuncToString): + (JSC::dateProtoFuncToUTCString): + (JSC::dateProtoFuncToISOString): + (JSC::dateProtoFuncToDateString): + (JSC::dateProtoFuncToTimeString): + (JSC::dateProtoFuncGetFullYear): + (JSC::dateProtoFuncGetUTCFullYear): + (JSC::dateProtoFuncToGMTString): + (JSC::dateProtoFuncGetMonth): + (JSC::dateProtoFuncGetUTCMonth): + (JSC::dateProtoFuncGetDate): + (JSC::dateProtoFuncGetUTCDate): + (JSC::dateProtoFuncGetDay): + (JSC::dateProtoFuncGetUTCDay): + (JSC::dateProtoFuncGetHours): + (JSC::dateProtoFuncGetUTCHours): + (JSC::dateProtoFuncGetMinutes): + (JSC::dateProtoFuncGetUTCMinutes): + (JSC::dateProtoFuncGetSeconds): + (JSC::dateProtoFuncGetUTCSeconds): + (JSC::dateProtoFuncGetTimezoneOffset): + (JSC::setNewValueFromTimeArgs): + (JSC::setNewValueFromDateArgs): + (JSC::dateProtoFuncSetYear): + (JSC::dateProtoFuncGetYear): Updated for the gregorianDateTime change above. + +2009-11-09 Geoffrey Garen <ggaren@apple.com> + + Build fix: export a new symbol. + + * JavaScriptCore.exp: + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + +2009-11-09 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Sam "Home Wrecker" Weinig. + + Added a tiny cache for Date parsing. + + SunSpider says 1.2% faster. + + * runtime/DateConversion.cpp: + (JSC::parseDate): Try to reuse the last parsed Date, if present. + + * runtime/JSGlobalData.cpp: + (JSC::JSGlobalData::resetDateCache): + * runtime/JSGlobalData.h: Added storage for last parsed Date. Refactored + this code to make resetting the date cache easier. + + * runtime/JSGlobalObject.h: + (JSC::DynamicGlobalObjectScope::DynamicGlobalObjectScope): Updated for + refactoring. + + * wtf/DateMath.cpp: + (JSC::parseDateFromNullTerminatedCharacters): + * wtf/DateMath.h: Changed ExecState to be first parameter, as is the JSC custom. + +2009-11-09 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + Can cache prototype lookups on uncacheable dictionaries. + https://bugs.webkit.org/show_bug.cgi?id=31198 + + Replace fromDictionaryTransition with flattenDictionaryObject and + flattenDictionaryStructure. This change is necessary as we need to + guarantee that our attempt to convert away from a dictionary structure + will definitely succeed, and in some cases this requires mutating the + object storage itself. + + * interpreter/Interpreter.cpp: + (JSC::Interpreter::tryCacheGetByID): + * jit/JITStubs.cpp: + (JSC::JITThunks::tryCacheGetByID): + (JSC::DEFINE_STUB_FUNCTION): + * runtime/BatchedTransitionOptimizer.h: + (JSC::BatchedTransitionOptimizer::~BatchedTransitionOptimizer): + * runtime/JSObject.h: + (JSC::JSObject::flattenDictionaryObject): + * runtime/Operations.h: + (JSC::normalizePrototypeChain): + * runtime/Structure.cpp: + (JSC::Structure::flattenDictionaryStructure): + (JSC::comparePropertyMapEntryIndices): + * runtime/Structure.h: + +2009-11-09 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Not reviewed, build fix. + + Remove extra character from r50701. + + * JavaScriptCore.pri: + +2009-11-09 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Not reviewed, build fix. + + Revert r50695 because it broke QtWebKit (clean builds). + + * JavaScriptCore.pri: + +2009-11-09 Norbert Leser <norbert.leser@nokia.com> + + Reviewed by Kenneth Rohde Christiansen. + + Prepended $$PWD to GENERATED_SOURCES_DIR to avoid potential ambiguities when included from WebCore.pro. + Some preprocessors consider this GENERATED_SOURCES_DIR relative to current invoking dir (e.g., ./WebCore), + and not the working dir of JavaCriptCore.pri (i.e., ../JavaScriptCore/). + + * JavaScriptCore.pri: + +2009-11-09 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Kenneth Rohde Christiansen. + + Use explicit parentheses to silence gcc 4.4 -Wparentheses warnings + https://bugs.webkit.org/show_bug.cgi?id=31040 + + * interpreter/Interpreter.cpp: + (JSC::Interpreter::privateExecute): + +2009-11-08 David Levin <levin@chromium.org> + + Reviewed by NOBODY (speculative snow leopard and windows build fixes). + + * wtf/DateMath.cpp: + (WTF::parseDateFromNullTerminatedCharacters): + (JSC::gregorianDateTimeToMS): + (JSC::msToGregorianDateTime): + (JSC::parseDateFromNullTerminatedCharacters): + * wtf/DateMath.h: + (JSC::GregorianDateTime::GregorianDateTime): + +2009-11-08 David Levin <levin@chromium.org> + + Reviewed by NOBODY (chromium build fix). + + Hopefully, the last build fix. + + Create better separation in DateMath about the JSC + and non-JSC portions. Also, only expose the non-JSC + version in the exports. + + * JavaScriptCore.exp: + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + * wtf/DateMath.cpp: + (WTF::parseDateFromNullTerminatedCharacters): + (JSC::getUTCOffset): + (JSC::gregorianDateTimeToMS): + (JSC::msToGregorianDateTime): + (JSC::parseDateFromNullTerminatedCharacters): + * wtf/DateMath.h: + (JSC::gmtoffset): + +2009-11-08 David Levin <levin@chromium.org> + + Reviewed by NOBODY (chromium build fix). + + For the change in DateMath. + + * config.h: + * wtf/DateMath.cpp: + +2009-11-06 Geoffrey Garen <ggaren@apple.com> + + Windows build fix: export some symbols. + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + +2009-11-06 Geoffrey Garen <ggaren@apple.com> + + Build fix: updated export file. + + * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def: + +2009-11-06 Geoffrey Garen <ggaren@apple.com> + + Build fix: added some #includes. + + * wtf/CurrentTime.h: + * wtf/DateMath.h: + +2009-11-06 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Oliver Hunt. + + https://bugs.webkit.org/show_bug.cgi?id=31197 + Implemented a timezone cache not based on Mac OS X's notify_check API. + + If the VM calculates the local timezone offset from UTC, it caches the + result until the end of the current VM invocation. (We don't want to cache + forever, because the user's timezone may change over time.) + + This removes notify_* overhead on Mac, and, more significantly, removes + OS time and date call overhead on non-Mac platforms. + + ~8% speedup on Date microbenchmark on Mac. SunSpider reports maybe a tiny + speedup on Mac. (Speedup on non-Mac platforms should be even more noticeable.) + + * JavaScriptCore.exp: + + * interpreter/CachedCall.h: + (JSC::CachedCall::CachedCall): + * interpreter/Interpreter.cpp: + (JSC::Interpreter::execute): + * runtime/JSGlobalObject.h: + (JSC::DynamicGlobalObjectScope::DynamicGlobalObjectScope): Made the + DynamicGlobalObjectScope constructor responsible for checking whether a + dynamicGlobalObject has already been set. This eliminated some duplicate + client code, and allowed me to avoid adding even more duplicate client + code. Made DynamicGlobalObjectScope responsible for resetting the + local timezone cache upon first entry to the VM. + + * runtime/DateConstructor.cpp: + (JSC::constructDate): + (JSC::callDate): + (JSC::dateParse): + (JSC::dateUTC): + * runtime/DateConversion.cpp: + (JSC::parseDate): + * runtime/DateConversion.h: + * runtime/DateInstance.cpp: + (JSC::DateInstance::gregorianDateTime): + * runtime/DateInstance.h: + * runtime/DateInstanceCache.h: + * runtime/DatePrototype.cpp: + (JSC::setNewValueFromTimeArgs): + (JSC::setNewValueFromDateArgs): + (JSC::dateProtoFuncSetYear): + * runtime/InitializeThreading.cpp: + (JSC::initializeThreadingOnce): + * runtime/JSGlobalData.cpp: + (JSC::JSGlobalData::JSGlobalData): + * runtime/JSGlobalData.h: + * wtf/DateMath.cpp: + (WTF::getCurrentUTCTime): + (WTF::getCurrentUTCTimeWithMicroseconds): + (WTF::getLocalTime): + (JSC::getUTCOffset): Use the new cache. Also, see below. + (JSC::gregorianDateTimeToMS): + (JSC::msToGregorianDateTime): + (JSC::initializeDates): + (JSC::parseDateFromNullTerminatedCharacters): Simplified the way this function + accounts for the local timezone offset, to accomodate our new caching API, + and a (possibly misguided) caller in WebCore. Also, see below. + * wtf/DateMath.h: + (JSC::GregorianDateTime::GregorianDateTime): Moved most of the code in + DateMath.* into the JSC namespace. The code needed to move so it could + naturally interact with ExecState and JSGlobalData to support caching. + Logically, it seemed right to move it, too, since this code is not really + as low-level as the WTF namespace might imply -- it implements a set of + date parsing and conversion quirks that are finely tuned to the JavaScript + language. Also removed the Mac OS X notify_* infrastructure. + + * wtf/CurrentTime.h: + (WTF::currentTimeMS): + (WTF::getLocalTime): Moved the rest of the DateMath code here, and renamed + it to make it consistent with WTF's currentTime function. + +2009-11-06 Gabor Loki <loki@inf.u-szeged.hu> + + Unreviewed trivial buildfix after r50595. + + Rename the remaining rshiftPtr calls to rshift32 + + * jit/JITArithmetic.cpp: + (JSC::JIT::emit_op_rshift): + * jit/JITInlineMethods.h: + (JSC::JIT::emitFastArithImmToInt): + +2009-11-06 Gavin Barraclough <barraclough@apple.com> + + Reviewed by Oliver Hunt. + + Tidy up the shift methods on the macro-assembler interface. + + Currently behaviour of shifts of a magnitude > 0x1f is undefined. + Instead defined that all shifts are masked to this range. This makes a lot of + practical sense, both since having undefined behaviour is not particularly + desirable, and because this behaviour is commonly required (particularly since + it is required bt ECMA-262 for shifts). + + Update the ARM assemblers to provide this behaviour. Remove (now) redundant + masks from JITArithmetic, and remove rshiftPtr (this was used in case that + could be rewritten in a simpler form using rshift32, only optimized JSVALUE32 + on x86-64, which uses JSVALUE64!) + + * assembler/MacroAssembler.h: + * assembler/MacroAssemblerARM.h: + (JSC::MacroAssemblerARM::lshift32): + (JSC::MacroAssemblerARM::rshift32): + * assembler/MacroAssemblerARMv7.h: + (JSC::MacroAssemblerARMv7::lshift32): + (JSC::MacroAssemblerARMv7::rshift32): + * assembler/MacroAssemblerX86_64.h: + * jit/JITArithmetic.cpp: + (JSC::JIT::emit_op_lshift): + (JSC::JIT::emit_op_rshift): + +2009-11-05 Gavin Barraclough <barraclough@apple.com> + + Rubber Stamped by Oliver Hunt. + + Remove a magic number (1) from the JIT, instead compute the value with OBJECT_OFFSET. + + * jit/JITInlineMethods.h: + (JSC::JIT::emitPutJITStubArg): + (JSC::JIT::emitPutJITStubArgConstant): + (JSC::JIT::emitGetJITStubArg): + (JSC::JIT::emitPutJITStubArgFromVirtualRegister): + * jit/JITStubCall.h: + (JSC::JITStubCall::JITStubCall): + (JSC::JITStubCall::getArgument): + * jit/JITStubs.h: + +2009-11-05 Zoltan Herczeg <zherczeg@inf.u-szeged.hu> + + Reviewed by Gavin Barraclough. + + https://bugs.webkit.org/show_bug.cgi?id=31159 + Fix branchDouble behaviour on ARM THUMB2 JIT. + + The x86 branchDouble behaviour is reworked, and all JIT + ports should follow the x86 port. See bug 31104 and 31151 + + This patch contains a fix for the traditional ARM port + + * assembler/ARMAssembler.h: + (JSC::ARMAssembler::): + (JSC::ARMAssembler::fmrs_r): + (JSC::ARMAssembler::ftosid_r): + * assembler/MacroAssemblerARM.h: + (JSC::MacroAssemblerARM::): + (JSC::MacroAssemblerARM::branchDouble): + (JSC::MacroAssemblerARM::branchConvertDoubleToInt32): + +2009-11-05 Chris Jerdonek <chris.jerdonek@gmail.com> + + Reviewed by Eric Seidel. + + Removed the "this is part of the KDE project" comments from + all *.h, *.cpp, *.idl, and *.pm files. + + https://bugs.webkit.org/show_bug.cgi?id=31167 + + The maintenance and architecture page in the project wiki lists + this as a task. + + This change includes no changes or additions to test cases + since the change affects only comments. + + * wtf/wince/FastMallocWince.h: + +2009-11-05 Gabor Loki <loki@inf.u-szeged.hu> + + Reviewed by Gavin Barraclough. + + Use ARMv7 specific encoding for immediate constants on ARMv7 target + https://bugs.webkit.org/show_bug.cgi?id=31060 + + * assembler/ARMAssembler.cpp: + (JSC::ARMAssembler::getOp2): Use INVALID_IMM + (JSC::ARMAssembler::getImm): Use encodeComplexImm for complex immediate + (JSC::ARMAssembler::moveImm): Ditto. + (JSC::ARMAssembler::encodeComplexImm): Encode a constant by one or two + instructions or a PC relative load. + * assembler/ARMAssembler.h: Use INVALID_IMM if a constant cannot be + encoded as an immediate constant. + (JSC::ARMAssembler::): + (JSC::ARMAssembler::movw_r): 16-bit immediate load + (JSC::ARMAssembler::movt_r): High halfword 16-bit immediate load + (JSC::ARMAssembler::getImm16Op2): Encode immediate constant for + movw_r and mowt_r + +2009-11-04 Mark Mentovai <mark@chromium.org> + + Reviewed by Mark Rowe. + + Provide TARGETING_TIGER and TARGETING_LEOPARD as analogues to + BUILDING_ON_TIGER and BUILDING_ON_LEOPARD. The TARGETING_ macros + consider the deployment target; the BUILDING_ON_ macros consider the + headers being built against. + + * wtf/Platform.h: + +2009-11-04 Gavin Barraclough <barraclough@apple.com> + + Reviewed by Oliver Hunt. + + https://bugs.webkit.org/show_bug.cgi?id=31151 + Fix branchDouble behaviour on ARM THUMB2 JIT. + + The ARMv7 JIT is currently using ARMv7Assembler::ConditionEQ to branch + for DoubleEqualOrUnordered, however this is incorrect – ConditionEQ won't + branch on unordered operands. Similarly, DoubleLessThanOrUnordered & + DoubleLessThanOrEqualOrUnordered use ARMv7Assembler::ConditionLO & + ARMv7Assembler::ConditionLS, whereas they should be using + ARMv7Assembler::ConditionLT & ARMv7Assembler::ConditionLE. + + Fix these, and fill out the missing DoubleConditions. + + * assembler/MacroAssemblerARMv7.h: + (JSC::MacroAssemblerARMv7::): + (JSC::MacroAssemblerARMv7::branchDouble): + +2009-11-04 Gavin Barraclough <barraclough@apple.com> + + Rubber Stamped by Oliver Hunt. + + Enable native call optimizations on ARMv7. (Existing ARM_TRADITIONAL + implementation was generic, worked perfectly, just needed turning on). + + * jit/JITOpcodes.cpp: + * wtf/Platform.h: + +2009-11-04 Gavin Barraclough <barraclough@apple.com> + + Rubber Stamped by Mark Rowe, Oliver Hunt, and Sam Weinig. + + Add a missing assert to the ARMv7 JIT. + + * assembler/ARMv7Assembler.h: + (JSC::ARMThumbImmediate::ARMThumbImmediate): + +2009-11-04 Mark Rowe <mrowe@apple.com> + + Rubber-stamped by Oliver Hunt. + + Remove bogus op_ prefix on dumped version of three opcodes. + + * bytecode/CodeBlock.cpp: + (JSC::CodeBlock::dump): + +2009-11-04 Mark Rowe <mrowe@apple.com> + + Reviewed by Sam Weinig. + + Fix dumping of constants in bytecode so that they aren't printed as large positive register numbers. + + We do this by having the registerName function return information about the constant if the register + number corresponds to a constant. This requires that registerName, and several functions that call it, + be converted to member functions of CodeBlock so that the constant value can be retrieved. The + ExecState also needs to be threaded down through these functions so that it can be passed on to + constantName when needed. + + * bytecode/CodeBlock.cpp: + (JSC::constantName): + (JSC::CodeBlock::registerName): + (JSC::CodeBlock::printUnaryOp): + (JSC::CodeBlock::printBinaryOp): + (JSC::CodeBlock::printConditionalJump): + (JSC::CodeBlock::printGetByIdOp): + (JSC::CodeBlock::printPutByIdOp): + (JSC::CodeBlock::dump): + * bytecode/CodeBlock.h: + (JSC::CodeBlock::isConstantRegisterIndex): + +2009-11-04 Pavel Heimlich <tropikhajma@gmail.com> + + Reviewed by Alexey Proskuryakov. + + https://bugs.webkit.org/show_bug.cgi?id=30647 + Solaris build failure due to strnstr. + + * wtf/StringExtras.h: Enable strnstr on Solaris, too. + +2009-11-04 Gavin Barraclough <barraclough@apple.com> + + Reviewed by Oliver Hunt. + + https://bugs.webkit.org/show_bug.cgi?id=31104 + Refactor x86-specific behaviour out of the JIT. + + - Add explicit double branch conditions for ordered and unordered comparisons (presently the brehaviour is a mix). + - Refactor double to int conversion out into the MacroAssembler. + - Remove broken double to int conversion for !JSVALUE32_64 builds - this code was broken and slowing us down, fixing it showed it not to be an improvement. + - Remove exclusion of double to int conversion from (1 % X) cases in JSVALUE32_64 builds - if this was of benefit this is no longer the case; simplify. + + * assembler/MacroAssemblerARM.h: + (JSC::MacroAssemblerARM::): + * assembler/MacroAssemblerARMv7.h: + (JSC::MacroAssemblerARMv7::): + * assembler/MacroAssemblerX86Common.h: + (JSC::MacroAssemblerX86Common::): + (JSC::MacroAssemblerX86Common::convertInt32ToDouble): + (JSC::MacroAssemblerX86Common::branchDouble): + (JSC::MacroAssemblerX86Common::branchConvertDoubleToInt32): + * jit/JITArithmetic.cpp: + (JSC::JIT::emitBinaryDoubleOp): + (JSC::JIT::emit_op_div): + (JSC::JIT::emitSlow_op_jnless): + (JSC::JIT::emitSlow_op_jnlesseq): + * jit/JITOpcodes.cpp: + (JSC::JIT::emit_op_jfalse): + +2009-11-04 Mark Mentovai <mark@chromium.org> + + Reviewed by Eric Seidel. + + Remove BUILDING_ON_LEOPARD from JavaScriptCore.gyp. This is supposed + to be set as needed only in wtf/Platform.h. + + * JavaScriptCore.gyp/JavaScriptCore.gyp: + +2009-11-02 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + REGRESSION (r48573): JSC may incorrectly cache chain lookups with a dictionary at the head of the chain + https://bugs.webkit.org/show_bug.cgi?id=31045 + + Add guards to prevent caching of prototype chain lookups with dictionaries at the + head of the chain. Also add a few tighter assertions to cached prototype lookups + to catch this in future. + + * interpreter/Interpreter.cpp: + (JSC::Interpreter::tryCacheGetByID): + (JSC::Interpreter::privateExecute): + * jit/JITStubs.cpp: + (JSC::JITThunks::tryCacheGetByID): + +2009-11-02 Laszlo Gombos <laszlo.1.gombos@nokia.com> + + Reviewed by Darin Adler. + + PLATFORM(CF) should be set when building for Qt on Darwin + https://bugs.webkit.org/show_bug.cgi?id=23671 + + * wtf/Platform.h: Turn on CF support if both QT and DARWIN + platforms are defined. + +2009-11-02 Dmitry Titov <dimich@chromium.org> + + Reviewed by David Levin. + + Remove threadsafe refcounting from tasks used with WTF::MessageQueue. + https://bugs.webkit.org/show_bug.cgi?id=30612 + + * wtf/MessageQueue.h: + (WTF::MessageQueue::alwaysTruePredicate): + (WTF::MessageQueue::~MessageQueue): + (WTF::MessageQueue::append): + (WTF::MessageQueue::appendAndCheckEmpty): + (WTF::MessageQueue::prepend): + (WTF::MessageQueue::waitForMessage): + (WTF::MessageQueue::waitForMessageFilteredWithTimeout): + (WTF::MessageQueue::tryGetMessage): + (WTF::MessageQueue::removeIf): + The MessageQueue is changed to act as a queue of OwnPtr<DataType>. It takes ownership + of posted tasks and passes it to the new owner (in another thread) when the task is fetched. + All methods have arguments of type PassOwnPtr<DataType> and return the same type. + + * wtf/Threading.cpp: + (WTF::createThread): + Superficial change to trigger rebuild of JSC project on Windows, + workaround for https://bugs.webkit.org/show_bug.cgi?id=30890 + +2009-10-30 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Oliver Hunt. + + Fixed failing layout test: restore a special case I accidentally deleted. + + * runtime/DatePrototype.cpp: + (JSC::setNewValueFromDateArgs): In the case of applying a change to a date + that is NaN, reset the date to 0 *and* then apply the change; don't just + reset the date to 0. + +2009-10-30 Geoffrey Garen <ggaren@apple.com> + + Windows build fix: update for object-to-pointer change. + + * runtime/DatePrototype.cpp: + (JSC::formatLocaleDate): + +2009-10-29 Geoffrey Garen <ggaren@apple.com> + + Reviewed by Darin Adler. + + https://bugs.webkit.org/show_bug.cgi?id=30942 + Use pointers instead of copies to pass GregorianDateTime objects around. + + SunSpider reports a shocking 4.5% speedup on date-format-xparb, and 1.3% + speedup on date-format-tofte. + + * runtime/DateInstance.cpp: + (JSC::DateInstance::gregorianDateTime): + * runtime/DateInstance.h: + * runtime/DatePrototype.cpp: + (JSC::formatLocaleDate): + (JSC::dateProtoFuncToString): + (JSC::dateProtoFuncToUTCString): + (JSC::dateProtoFuncToISOString): + (JSC::dateProtoFuncToDateString): + (JSC::dateProtoFuncToTimeString): + (JSC::dateProtoFuncGetFullYear): + (JSC::dateProtoFuncGetUTCFullYear): + (JSC::dateProtoFuncToGMTString): + (JSC::dateProtoFuncGetMonth): + (JSC::dateProtoFuncGetUTCMonth): + (JSC::dateProtoFuncGetDate): + (JSC::dateProtoFuncGetUTCDate): + (JSC::dateProtoFuncGetDay): + (JSC::dateProtoFuncGetUTCDay): + (JSC::dateProtoFuncGetHours): + (JSC::dateProtoFuncGetUTCHours): + (JSC::dateProtoFuncGetMinutes): + (JSC::dateProtoFuncGetUTCMinutes): + (JSC::dateProtoFuncGetSeconds): + (JSC::dateProtoFuncGetUTCSeconds): + (JSC::dateProtoFuncGetTimezoneOffset): + (JSC::setNewValueFromTimeArgs): + (JSC::setNewValueFromDateArgs): + (JSC::dateProtoFuncSetYear): + (JSC::dateProtoFuncGetYear): Renamed getGregorianDateTime to gregorianDateTime, + since it no longer has an out parameter. Uses 0 to indicate invalid dates. + +2009-10-30 Zoltan Horvath <zoltan@webkit.org> + + Reviewed by Darin Adler. + + Allow custom memory allocation control for JavaScriptCore's ListHashSet + https://bugs.webkit.org/show_bug.cgi?id=30853 + + Inherits ListHashSet class from FastAllocBase because it is + instantiated by 'new' in WebCore/rendering/RenderBlock.cpp:1813. + + * wtf/ListHashSet.h: + +2009-10-30 Oliver Hunt <oliver@apple.com> + + Reviewed by Gavin Barraclough. + + Regression: crash enumerating properties of an object with getters or setters + https://bugs.webkit.org/show_bug.cgi?id=30948 + + Add a guard to prevent us trying to cache property enumeration on + objects with getters or setters. + + * runtime/JSPropertyNameIterator.cpp: + (JSC::JSPropertyNameIterator::create): + +2009-10-30 Roland Steiner <rolandsteiner@chromium.org> + + Reviewed by Eric Seidel. + + Remove ENABLE_RUBY guards as discussed with Dave Hyatt and Maciej Stachowiak. + + Bug 28420 - Implement HTML5 <ruby> rendering + (https://bugs.webkit.org/show_bug.cgi?id=28420) + + No new tests (no functional change). + + * Configurations/FeatureDefines.xcconfig: + +2009-10-29 Oliver Hunt <oliver@apple.com> + + Reviewed by Maciej Stachowiak. + + REGRESSION (r50218-r50262): E*TRADE accounts page is missing content + https://bugs.webkit.org/show_bug.cgi?id=30947 + <rdar://problem/7348833> + + The logic for flagging that a structure has non-enumerable properties + was in addPropertyWithoutTransition, rather than in the core Structure::put + method. Despite this I was unable to produce a testcase that caused + the failure that etrade was experiencing, but the new assertion in + getEnumerablePropertyNames triggers on numerous layout tests without + the fix, so in effecti all for..in enumeration in any test ends up + doing the required consistency check. + + * runtime/Structure.cpp: + (JSC::Structure::addPropertyWithoutTransition): + (JSC::Structure::put): + (JSC::Structure::getEnumerablePropertyNames): + (JSC::Structure::checkConsistency): + +2009-10-29 Gabor Loki <loki@inf.u-szeged.hu> + + Reviewed by Gavin Barraclough. + + Add cacheFlush support for Thumb-2 on Linux + https://bugs.webkit.org/show_bug.cgi?id=30865 + + * jit/ExecutableAllocator.h: + (JSC::ExecutableAllocator::cacheFlush): + 2009-10-28 Gavin Barraclough <barraclough@apple.com> Reviewed by Oliver Hunt. |