[JSC] Add basic DFG/FTL support for Math.round
authorbenjamin@webkit.org <benjamin@webkit.org@268f45cc-cd09-0410-ab3c-d52691b4dbfc>
Fri, 8 May 2015 00:23:32 +0000 (00:23 +0000)
committerbenjamin@webkit.org <benjamin@webkit.org@268f45cc-cd09-0410-ab3c-d52691b4dbfc>
Fri, 8 May 2015 00:23:32 +0000 (00:23 +0000)
commite248f0fd7dc4f7160708aa06fe3028010a7039ab
tree8afa64d2e71ba9ce5f9e1909044299377969a0c4
parent719739c22f611ed7c83ed7a8a29d7cd8166faece
[JSC] Add basic DFG/FTL support for Math.round
https://bugs.webkit.org/show_bug.cgi?id=144725

Patch by Benjamin Poulain <bpoulain@apple.com> on 2015-05-07
Reviewed by Filip Pizlo.

This patch adds two optimizations targeting Math.round():
-Add a DFGNode ArithRound corresponding to the intrinsic RoundIntrinsic.
-Change the MacroAssembler to be stricter on how we fail to convert a double
 to ingeter. Previously, any number valued zero would fail, now we only
 fail for -0.

Since ArithRound speculate it produces int32, the MacroAssembler assembler
part became necessary because zero is a pretty common output of Math.round()
and we would OSR exit a lot (and eventually recompile for doubles).

The implementation itself of the inline Math.round() is exactly the same
as the C function that exists for Math.round(). We can very likely do better
but it is a good start known to be valid and inlining alone alread provides
significant speedups.

* assembler/X86Assembler.h:
(JSC::X86Assembler::movmskpd_rr):
* assembler/MacroAssemblerX86Common.h:
(JSC::MacroAssemblerX86Common::branchConvertDoubleToInt32):
When we have a zero, get the sign bit out of the double and check if is one.

I'll look into doing the same improvement for ARM.

* bytecode/SpeculatedType.cpp:
(JSC::typeOfDoubleRounding):
(JSC::typeOfDoubleFRound): Deleted.
* bytecode/SpeculatedType.h:
* dfg/DFGAbstractInterpreterInlines.h:
(JSC::DFG::AbstractInterpreter<AbstractStateType>::executeEffects):
* dfg/DFGByteCodeParser.cpp:
(JSC::DFG::ByteCodeParser::handleIntrinsic):
* dfg/DFGClobberize.h:
(JSC::DFG::clobberize):
* dfg/DFGDoesGC.cpp:
(JSC::DFG::doesGC):
* dfg/DFGFixupPhase.cpp:
(JSC::DFG::FixupPhase::fixupNode):
* dfg/DFGGraph.h:
(JSC::DFG::Graph::roundShouldSpeculateInt32):
(JSC::DFG::Graph::negateShouldSpeculateMachineInt): Deleted.
* dfg/DFGNode.h:
(JSC::DFG::Node::arithNodeFlags):
(JSC::DFG::Node::hasHeapPrediction):
(JSC::DFG::Node::hasArithMode):
* dfg/DFGNodeType.h:
* dfg/DFGPredictionPropagationPhase.cpp:
(JSC::DFG::PredictionPropagationPhase::propagate):
* dfg/DFGSafeToExecute.h:
(JSC::DFG::safeToExecute):
* dfg/DFGSpeculativeJIT.cpp:
(JSC::DFG::SpeculativeJIT::compileArithRound):
* dfg/DFGSpeculativeJIT.h:
* dfg/DFGSpeculativeJIT32_64.cpp:
(JSC::DFG::SpeculativeJIT::compile):
* dfg/DFGSpeculativeJIT64.cpp:
(JSC::DFG::SpeculativeJIT::compile):
* ftl/FTLCapabilities.cpp:
(JSC::FTL::canCompile):
* ftl/FTLIntrinsicRepository.h:
* ftl/FTLLowerDFGToLLVM.cpp:
(JSC::FTL::LowerDFGToLLVM::compileNode):
(JSC::FTL::LowerDFGToLLVM::convertDoubleToInt32):
(JSC::FTL::LowerDFGToLLVM::compileDoubleAsInt32):
(JSC::FTL::LowerDFGToLLVM::compileArithRound):
* ftl/FTLOutput.h:
(JSC::FTL::Output::ceil64):
* jit/ThunkGenerators.cpp:
* runtime/MathCommon.cpp:
* runtime/MathCommon.h:
* runtime/MathObject.cpp:
(JSC::mathProtoFuncRound):
* tests/stress/math-round-basics.js: Added.
(mathRoundOnIntegers):
(mathRoundOnDoubles):
(mathRoundOnBooleans):
(uselessMathRound):
(mathRoundWithOverflow):
(mathRoundConsumedAsDouble):
(mathRoundDoesNotCareAboutMinusZero):
(mathRoundNoArguments):
(mathRoundTooManyArguments):
(testMathRoundOnConstants):
(mathRoundStructTransition):
(Math.round):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@183963 268f45cc-cd09-0410-ab3c-d52691b4dbfc
29 files changed:
Source/JavaScriptCore/ChangeLog
Source/JavaScriptCore/assembler/MacroAssemblerX86Common.h
Source/JavaScriptCore/assembler/X86Assembler.h
Source/JavaScriptCore/bytecode/SpeculatedType.cpp
Source/JavaScriptCore/bytecode/SpeculatedType.h
Source/JavaScriptCore/dfg/DFGAbstractInterpreterInlines.h
Source/JavaScriptCore/dfg/DFGArithMode.h
Source/JavaScriptCore/dfg/DFGByteCodeParser.cpp
Source/JavaScriptCore/dfg/DFGClobberize.h
Source/JavaScriptCore/dfg/DFGDoesGC.cpp
Source/JavaScriptCore/dfg/DFGFixupPhase.cpp
Source/JavaScriptCore/dfg/DFGGraph.h
Source/JavaScriptCore/dfg/DFGNode.h
Source/JavaScriptCore/dfg/DFGNodeType.h
Source/JavaScriptCore/dfg/DFGPredictionPropagationPhase.cpp
Source/JavaScriptCore/dfg/DFGSafeToExecute.h
Source/JavaScriptCore/dfg/DFGSpeculativeJIT.cpp
Source/JavaScriptCore/dfg/DFGSpeculativeJIT.h
Source/JavaScriptCore/dfg/DFGSpeculativeJIT32_64.cpp
Source/JavaScriptCore/dfg/DFGSpeculativeJIT64.cpp
Source/JavaScriptCore/ftl/FTLCapabilities.cpp
Source/JavaScriptCore/ftl/FTLIntrinsicRepository.h
Source/JavaScriptCore/ftl/FTLLowerDFGToLLVM.cpp
Source/JavaScriptCore/ftl/FTLOutput.h
Source/JavaScriptCore/jit/ThunkGenerators.cpp
Source/JavaScriptCore/runtime/MathCommon.cpp
Source/JavaScriptCore/runtime/MathCommon.h
Source/JavaScriptCore/runtime/MathObject.cpp
Source/JavaScriptCore/tests/stress/math-round-basics.js [new file with mode: 0644]