Vm-dev March 2017

vm-dev@lists.squeakfoundation.org

39 participants
154 discussions

Call for big benchmarks
by Javier Pimás 25 Mar '17

25 Mar '17

Hi everybody! While measuring performance I usually face the problem of assessing performance. At present, I'm using are-we-fast-yet benchs and also smark ones, among others. Unfortunately, for some purposes those benchs are considered too low-level, so I'd like to collect a set of "real world" fat workloads. I'm asking for contributions on computing problems you have. I'll create a public repo as a common place, explaining what each bench does, so that we can all benefit from the result. The code should have the following properties: - The executed code should be > 20KLOC. - It should have compute or memory intensive models or both. - Ideally, it has sufficient run time, perhaps a few minutes, but at least a few seconds. - The results should be verifiable. - The more portable the best, as I need to have those benchs running in another dialect. - Should be fully automate-able, it should be possible to suppress output. Some examples are moose models, source code analysis, parallelization, web apps, parser generators, their corresponding parsers, XML processing, report generation, refactoring tools, type inference, etc… Feel free to forward this mail to other lists you consider appropriate. Cheers, Pocho -- Javier Pimás Ciudad de Buenos Aires

6 13

[OpenSmalltalk/opensmalltalk-vm] fa67b2: CogVM source as per VMMaker.oscog-eem.2183
by GitHub 25 Mar '17

25 Mar '17

Branch: refs/heads/Cog Home: https://github.com/OpenSmalltalk/opensmalltalk-vm Commit: fa67b2b0b225165b427da4217601c457b680363d https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/fa67b2b0b225165b42… Author: Eliot Miranda <eliot.miranda(a)gmail.com> Date: 2017-03-24 (Fri, 24 Mar 2017) Changed paths: M nsspur64src/vm/cogit.h M nsspur64src/vm/cointerp.c M nsspur64src/vm/cointerp.h M nsspur64src/vm/gcc3x-cointerp.c M nsspursrc/vm/cogit.h M nsspursrc/vm/cointerp.c M nsspursrc/vm/cointerp.h M nsspursrc/vm/gcc3x-cointerp.c M nsspurstack64src/vm/gcc3x-interp.c M nsspurstack64src/vm/interp.c M nsspurstacksrc/vm/gcc3x-interp.c M nsspurstacksrc/vm/interp.c M spur64src/vm/cogit.h M spur64src/vm/cointerp.c M spur64src/vm/cointerp.h M spur64src/vm/gcc3x-cointerp.c M spurlowcode64src/vm/cogit.h M spurlowcode64src/vm/cointerp.c M spurlowcode64src/vm/cointerp.h M spurlowcode64src/vm/gcc3x-cointerp.c M spurlowcodesrc/vm/cogit.h M spurlowcodesrc/vm/cointerp.c M spurlowcodesrc/vm/cointerp.h M spurlowcodesrc/vm/gcc3x-cointerp.c M spurlowcodestack64src/vm/gcc3x-interp.c M spurlowcodestack64src/vm/interp.c M spurlowcodestacksrc/vm/gcc3x-interp.c M spurlowcodestacksrc/vm/interp.c M spursista64src/vm/cogit.h M spursista64src/vm/cointerp.c M spursista64src/vm/cointerp.h M spursista64src/vm/gcc3x-cointerp.c M spursistasrc/vm/cogit.h M spursistasrc/vm/cointerp.c M spursistasrc/vm/cointerp.h M spursistasrc/vm/gcc3x-cointerp.c M spursrc/vm/cogit.h M spursrc/vm/cointerp.c M spursrc/vm/cointerp.h M spursrc/vm/gcc3x-cointerp.c M spurstack64src/vm/gcc3x-interp.c M spurstack64src/vm/interp.c M spurstacksrc/vm/gcc3x-interp.c M spurstacksrc/vm/interp.c M src/vm/cogit.h M src/vm/cointerp.c M src/vm/cointerp.h M src/vm/cointerpmt.c M src/vm/cointerpmt.h M src/vm/gcc3x-cointerp.c M src/vm/gcc3x-cointerpmt.c M stacksrc/vm/gcc3x-interp.c M stacksrc/vm/interp.c Log Message: ----------- CogVM source as per VMMaker.oscog-eem.2183 Fix the new validBCPC:inMethod: to use a sensible way of determining the initialPC.

1 0

VM Maker: VMMaker.oscog-eem.2183.mcz
by commits＠source.squeak.org 25 Mar '17

25 Mar '17

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.2183.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2183 Author: eem Time: 24 March 2017, 5:27:02.066951 pm UUID: 737e7483-f203-4279-afe5-d8f28e3d3f46 Ancestors: VMMaker.oscog-eem.2182 Fix the new validBCPC:inMethod: to use a sensible way of determining the initialPC. =============== Diff against VMMaker.oscog-eem.2182 =============== Item was changed: ----- Method: StackInterpreter>>validBCPC:inMethod: (in category 'debug support') ----- validBCPC: thePC inMethod: aMethod <var: #aMethod type: #usqInt> "Note that we accept anInstrPointer pointing to a callPrimitiveBytecode at the start of a method that contains a primitive. This because methods like Context(Part)>>reset have to be updated to skip the callPrimtiive bytecode otherwise." + ^thePC >= (self startPCOfMethod: aMethod) - "-1 for pre-increment in fetchNextBytecode" - ^thePC >= (objectMemory lastPointerOf: aMethod) and: [thePC < (objectMemory numBytesOfBytes: aMethod)]!

1 0

[OpenSmalltalk/opensmalltalk-vm] e315ac: CogVM source as per VMMaker.oscog-eem.2182
by GitHub 25 Mar '17

25 Mar '17

Branch: refs/heads/Cog Home: https://github.com/OpenSmalltalk/opensmalltalk-vm Commit: e315acf0ca9bf564dda57f57b2ab653f1287a43d https://github.com/OpenSmalltalk/opensmalltalk-vm/commit/e315acf0ca9bf564dd… Author: Eliot Miranda <eliot.miranda(a)gmail.com> Date: 2017-03-24 (Fri, 24 Mar 2017) Changed paths: M nsspur64src/vm/cogit.h M nsspur64src/vm/cogitX64SysV.c M nsspur64src/vm/cogitX64WIN64.c M nsspur64src/vm/cointerp.c M nsspur64src/vm/cointerp.h M nsspur64src/vm/gcc3x-cointerp.c M nsspursrc/vm/cogit.h M nsspursrc/vm/cogitARMv5.c M nsspursrc/vm/cogitIA32.c M nsspursrc/vm/cogitMIPSEL.c M nsspursrc/vm/cointerp.c M nsspursrc/vm/cointerp.h M nsspursrc/vm/gcc3x-cointerp.c M nsspurstack64src/vm/gcc3x-interp.c M nsspurstack64src/vm/interp.c M nsspurstacksrc/vm/gcc3x-interp.c M nsspurstacksrc/vm/interp.c M platforms/iOS/vm/Common/Classes/sqMacV2Time.c M spur64src/vm/cogit.h M spur64src/vm/cogitX64SysV.c M spur64src/vm/cogitX64WIN64.c M spur64src/vm/cointerp.c M spur64src/vm/cointerp.h M spur64src/vm/gcc3x-cointerp.c M spurlowcode64src/vm/cogit.h M spurlowcode64src/vm/cogitX64SysV.c M spurlowcode64src/vm/cogitX64WIN64.c M spurlowcode64src/vm/cointerp.c M spurlowcode64src/vm/cointerp.h M spurlowcode64src/vm/gcc3x-cointerp.c M spurlowcodesrc/vm/cogit.h M spurlowcodesrc/vm/cogitARMv5.c M spurlowcodesrc/vm/cogitIA32.c M spurlowcodesrc/vm/cogitMIPSEL.c M spurlowcodesrc/vm/cointerp.c M spurlowcodesrc/vm/cointerp.h M spurlowcodesrc/vm/gcc3x-cointerp.c M spurlowcodestack64src/vm/gcc3x-interp.c M spurlowcodestack64src/vm/interp.c M spurlowcodestacksrc/vm/gcc3x-interp.c M spurlowcodestacksrc/vm/interp.c M spursista64src/vm/cogit.h M spursista64src/vm/cogitX64SysV.c M spursista64src/vm/cogitX64WIN64.c M spursista64src/vm/cointerp.c M spursista64src/vm/cointerp.h M spursista64src/vm/gcc3x-cointerp.c M spursistasrc/vm/cogit.h M spursistasrc/vm/cogitARMv5.c M spursistasrc/vm/cogitIA32.c M spursistasrc/vm/cogitMIPSEL.c M spursistasrc/vm/cointerp.c M spursistasrc/vm/cointerp.h M spursistasrc/vm/gcc3x-cointerp.c M spursrc/vm/cogit.h M spursrc/vm/cogitARMv5.c M spursrc/vm/cogitIA32.c M spursrc/vm/cogitMIPSEL.c M spursrc/vm/cointerp.c M spursrc/vm/cointerp.h M spursrc/vm/gcc3x-cointerp.c M spurstack64src/vm/gcc3x-interp.c M spurstack64src/vm/interp.c M spurstacksrc/vm/gcc3x-interp.c M spurstacksrc/vm/interp.c M src/vm/cogit.h M src/vm/cogitARMv5.c M src/vm/cogitIA32.c M src/vm/cogitMIPSEL.c M src/vm/cointerp.c M src/vm/cointerp.h M src/vm/cointerpmt.c M src/vm/cointerpmt.h M src/vm/gcc3x-cointerp.c M src/vm/gcc3x-cointerpmt.c M stacksrc/vm/gcc3x-interp.c M stacksrc/vm/interp.c Log Message: ----------- CogVM source as per VMMaker.oscog-eem.2182 Fix pc mapping for backward jumps in Cogit VMs with extension bytecodes. The old code mapped a backward branch mcpc to the unextended jump bytecode, which could turn a backward branch into a forward jump, with disastrous results. The map searcher must offset by the number of extensions to refer to the beginning of the extended bytecode sequence for a backward jump. The extended backward jump bytecode must compute correctly the offset of the start of the extended bytecode from the target of the jump. Fix context printing so that the cogMethod is printed last in a stable context. Sista Cogit: Added and fixed a quite stable version of inlinedByteEqual. Added experimental version of fillArray. Temporary fix for inlined new. Fix lowCode primitives in Sista VM.

1 0

VM Maker: VMMaker.oscog-eem.2182.mcz
by commits＠source.squeak.org 25 Mar '17

25 Mar '17

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.2182.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2182 Author: eem Time: 24 March 2017, 4:03:46.057141 pm UUID: 541d6c6d-ade5-47f1-ad2b-f9278422e3c9 Ancestors: VMMaker.oscog-eem.2181 Fix pc mapping for backward jumps in Cogit VMs with extension bytecodes. The old code mapped a backward branch mcpc to the unextended jump bytecode, which could turn a backward branch into a forward jump, with disastrous results. The map searcher must offset by the number of extensions to refer to the beginning of the extended bytecode sequence for a backward jump. The extended backward jump bytecode must compute correctly the offset of the start of the extended bytecode from the target of the jump. Fix context printing so that the cogMethod is printed last in a stable context (I get confused otherwise). =============== Diff against VMMaker.oscog-eem.2181 =============== Item was changed: ----- Method: CoInterpreter>>ensureContextHasBytecodePC: (in category 'frame access') ----- ensureContextHasBytecodePC: aContext "Make sure the context has a byetcode pc. Can only be used on single contexts." | pc | self assert: (self isMarriedOrWidowedContext: aContext) not. pc := objectMemory fetchPointer: InstructionPointerIndex ofObject: aContext. ((objectMemory isIntegerObject: pc) and: [(pc := objectMemory integerValueOf: pc) < 0]) ifTrue: [pc := self mustMapMachineCodePC: pc context: aContext. + self assert: (self validBCPC: (objectMemory integerValueOf: pc) inMethod: (objectMemory fetchPointer: MethodIndex ofObject: aContext)). objectMemory storePointerUnchecked: InstructionPointerIndex ofObject: aContext withValue: pc]! Item was changed: ----- Method: CoInterpreter>>ifBackwardsCheckForEvents: (in category 'jump bytecodes') ----- + ifBackwardsCheckForEvents: offsetToJumpBytecode - ifBackwardsCheckForEvents: offset "Backward jump means we're in a loop. - check for possible interrupts. - check for long-running loops and JIT if appropriate." | switched backwardJumpCountByte | <inline: true> + offsetToJumpBytecode >= 0 ifTrue: - offset >= 0 ifTrue: [^self]. localSP < stackLimit ifTrue: [self externalizeIPandSP. switched := self checkForEventsMayContextSwitch: true. self returnToExecutive: true postContextSwitch: switched. self browserPluginReturnIfNeeded. self internalizeIPandSP. switched ifTrue: [^self]]. "We use the least significant byte of the flags word (which is marked as an immediate) and subtract two each time to avoid disturbing the least significant tag bit. Since the byte is initialized to 1 (on frame build), on first decrement it will become -1. Trip when it reaches 1 again." backwardJumpCountByte := self iframeBackwardBranchByte: localFP. (backwardJumpCountByte := backwardJumpCountByte - 2) = 1 ifTrue: [(self methodWithHeaderShouldBeCogged: (objectMemory methodHeaderOf: method)) ifTrue: [self externalizeIPandSP. + self attemptToSwitchToMachineCode: (self oopForPointer: localIP) - offsetToJumpBytecode - method - objectMemory baseHeaderSize - 1 - self attemptToSwitchToMachineCode: (self oopForPointer: localIP) - offset - method - objectMemory baseHeaderSize - 1 "If attemptToSwitchToMachineCode: returns the method could not be cogged, hence..."]. "can't cog method; avoid asking to cog it again for the longest possible time." backwardJumpCountByte := 16r7F] ifFalse: [backwardJumpCountByte = -1 ifTrue: "initialize the count" [self assert: minBackwardJumpCountForCompile <= 128. backwardJumpCountByte := minBackwardJumpCountForCompile - 1 << 1 + 1]]. self iframeBackwardBranchByte: localFP put: backwardJumpCountByte! Item was changed: ----- Method: CoInterpreter>>printMethodFieldForPrintContext: (in category 'debug printing') ----- printMethodFieldForPrintContext: aContext <inline: true> | meth | meth := objectMemory fetchPointer: MethodIndex ofObject: aContext. + (self isMarriedOrWidowedContext: aContext) + ifFalse: + [self shortPrintOop: meth. + (self methodHasCogMethod: meth) ifTrue: + [self space; printHexnp: (self cogMethodOf: meth)]] + ifTrue: + [(self methodHasCogMethod: meth) ifTrue: + [self printHexnp: (self cogMethodOf: meth); space]. + self shortPrintOop: meth]! - (self methodHasCogMethod: meth) ifTrue: - [self printHexnp: (self cogMethodOf: meth); space]. - self shortPrintOop: meth.! Item was added: + ----- Method: CogVMSimulator>>extUnconditionalJump (in category 'jump bytecodes') ----- + extUnconditionalJump + "242 11110010 i i i i i i i i Jump i i i i i i i i (+ Extend B * 256, where bbbbbbbb = sddddddd, e.g. -32768 = i=0, a=0, s=1)" + "| byte offset | + byte := objectMemory byteAt: localIP + 1. + offset := byte + (extB << 8). + (offset < 0 and: [(self iframeBackwardBranchByte: localFP) - 2 = 1]) ifTrue: [self halt]." + ^super extUnconditionalJump! Item was added: + ----- Method: CogVMSimulator>>primitiveVoidVMState (in category 'system control primitives') ----- + primitiveVoidVMState + self halt. + ^super primitiveVoidVMState! Item was changed: ----- Method: Cogit>>mapFor:bcpc:performUntil:arg: (in category 'method map') ----- mapFor: cogMethod bcpc: startbcpc performUntil: functionSymbol arg: arg "Machine-code <-> bytecode pc mapping support. Evaluate functionSymbol for each mcpc, bcpc pair in the map until the function returns non-zero, answering that result, or 0 if it fails to. To cut down on number of arguments. and to be usable for both pc-mapping and method introspection, we encode the annotation and the isBackwardBranch flag in the same parameter. Guilty as charged." <var: #cogMethod type: #'CogBlockMethod *'> <var: #functionSymbol declareC: 'sqInt (*functionSymbol)(BytecodeDescriptor *desc, sqInt annotationAndIsBackwardBranch, char *mcpc, sqInt bcpc, void *arg)'> <var: #arg type: #'void *'> <inline: true> | isInBlock mcpc bcpc endbcpc map mapByte homeMethod aMethodObj result latestContinuation byte descriptor bsOffset nExts annotation | <var: #descriptor type: #'BytecodeDescriptor *'> <var: #homeMethod type: #'CogMethod *'> self assert: cogMethod stackCheckOffset > 0. mcpc := cogMethod asUnsignedInteger + cogMethod stackCheckOffset. "The stack check maps to the start of the first bytecode, the first bytecode being effectively after frame build." result := self perform: functionSymbol with: nil with: 0 + (HasBytecodePC << 1) with: (self cCoerceSimple: mcpc to: #'char *') with: startbcpc with: arg. result ~= 0 ifTrue: [^result]. bcpc := startbcpc. "In both CMMethod and CMBlock cases find the start of the map and skip forward to the bytecode pc map entry for the stack check." cogMethod cmType = CMMethod ifTrue: [isInBlock := cogMethod cmIsFullBlock. homeMethod := self cCoerceSimple: cogMethod to: #'CogMethod *'. self assert: startbcpc = (coInterpreter startPCOfMethodHeader: homeMethod methodHeader). map := self mapStartFor: homeMethod. annotation := (objectMemory byteAt: map) >> AnnotationShift. self assert: (annotation = IsAbsPCReference or: [annotation = IsObjectReference or: [annotation = IsRelativeCall or: [annotation = IsDisplacementX2N]]]). latestContinuation := startbcpc. aMethodObj := homeMethod methodObject. endbcpc := (objectMemory numBytesOf: aMethodObj) - 1. bsOffset := self bytecodeSetOffsetForHeader: homeMethod methodHeader. "If the method has a primitive, skip it and the error code store, if any; Logically. these come before the stack check and so must be ignored." bcpc := bcpc + (self deltaToSkipPrimAndErrorStoreIn: aMethodObj header: homeMethod methodHeader)] ifFalse: [isInBlock := true. self assert: bcpc = cogMethod startpc. homeMethod := cogMethod cmHomeMethod. map := self findMapLocationForMcpc: cogMethod asUnsignedInteger + (self sizeof: CogBlockMethod) inMethod: homeMethod. self assert: map ~= 0. annotation := (objectMemory byteAt: map) >> AnnotationShift. self assert: (annotation >> AnnotationShift = HasBytecodePC "fiducial" or: [annotation >> AnnotationShift = IsDisplacementX2N]). [(annotation := (objectMemory byteAt: map) >> AnnotationShift) ~= HasBytecodePC] whileTrue: [map := map - 1]. map := map - 1. "skip fiducial; i.e. the map entry for the pc immediately following the method header." aMethodObj := homeMethod methodObject. bcpc := startbcpc - (self blockCreationBytecodeSizeForHeader: homeMethod methodHeader). bsOffset := self bytecodeSetOffsetForHeader: homeMethod methodHeader. byte := (objectMemory fetchByte: bcpc ofObject: aMethodObj) + bsOffset. descriptor := self generatorAt: byte. endbcpc := self nextBytecodePCFor: descriptor at: bcpc exts: -1 in: aMethodObj. bcpc := startbcpc]. nExts := 0. self inlineCacheTagsAreIndexes ifTrue: [enumeratingCogMethod := homeMethod]. "Now skip up through the bytecode pc map entry for the stack check." [(objectMemory byteAt: map) >> AnnotationShift ~= HasBytecodePC] whileTrue: [map := map - 1]. map := map - 1. [(mapByte := objectMemory byteAt: map) ~= MapEnd] whileTrue: "defensive; we exit on bcpc" [mapByte >= FirstAnnotation ifTrue: [| nextBcpc isBackwardBranch | annotation := mapByte >> AnnotationShift. mcpc := mcpc + ((mapByte bitAnd: DisplacementMask) * backEnd codeGranularity). (self isPCMappedAnnotation: annotation) ifTrue: [(annotation = IsSendCall and: [(mapByte := objectMemory byteAt: map - 1) >> AnnotationShift = IsAnnotationExtension]) ifTrue: [annotation := annotation + (mapByte bitAnd: DisplacementMask). map := map - 1]. [byte := (objectMemory fetchByte: bcpc ofObject: aMethodObj) + bsOffset. descriptor := self generatorAt: byte. isInBlock ifTrue: [bcpc >= endbcpc ifTrue: [^0]] ifFalse: [(descriptor isReturn and: [bcpc >= latestContinuation]) ifTrue: [^0]. (descriptor isBranch or: [descriptor isBlockCreation]) ifTrue: [| targetPC | targetPC := self latestContinuationPCFor: descriptor at: bcpc exts: nExts in: aMethodObj. latestContinuation := latestContinuation max: targetPC]]. nextBcpc := self nextBytecodePCFor: descriptor at: bcpc exts: nExts in: aMethodObj. descriptor isMapped or: [isInBlock and: [descriptor isMappedInBlock]]] whileFalse: [bcpc := nextBcpc. nExts := descriptor isExtension ifTrue: [nExts + 1] ifFalse: [0]]. isBackwardBranch := descriptor isBranch and: [self isBackwardBranch: descriptor at: bcpc exts: nExts in: aMethodObj]. result := self perform: functionSymbol with: descriptor with: (isBackwardBranch ifTrue: [annotation << 1 + 1] ifFalse: [annotation << 1]) with: (self cCoerceSimple: mcpc to: #'char *') + with: (isBackwardBranch ifTrue: [bcpc - (2 * nExts)] ifFalse: [bcpc]) - with: bcpc with: arg. result ~= 0 ifTrue: [^result]. bcpc := nextBcpc. nExts := descriptor isExtension ifTrue: [nExts + 1] ifFalse: [0]]] ifFalse: [self assert: (mapByte >> AnnotationShift = IsDisplacementX2N or: [mapByte >> AnnotationShift = IsAnnotationExtension]). mapByte < (IsAnnotationExtension << AnnotationShift) ifTrue: [mcpc := mcpc + ((mapByte - DisplacementX2N << AnnotationShift) * backEnd codeGranularity)]]. map := map - 1]. ^0! Item was changed: ----- Method: StackInterpreter>>extUnconditionalJump (in category 'jump bytecodes') ----- extUnconditionalJump "242 11110010 i i i i i i i i Jump i i i i i i i i (+ Extend B * 256, where bbbbbbbb = sddddddd, e.g. -32768 = i=0, a=0, s=1)" + | byte offset bcpcDelta | - | byte offset | byte := self fetchByte. offset := byte + (extB << 8). + bcpcDelta := offset < 0 ifTrue: [numExtB * 2] ifFalse: [0]. extB := 0. numExtB := 0. localIP := localIP + offset. + self ifBackwardsCheckForEvents: offset + bcpcDelta. - self ifBackwardsCheckForEvents: offset. self fetchNextBytecode! Item was added: + ----- Method: StackInterpreter>>validBCPC:inMethod: (in category 'debug support') ----- + validBCPC: thePC inMethod: aMethod + <var: #aMethod type: #usqInt> + "Note that we accept anInstrPointer pointing to a callPrimitiveBytecode + at the start of a method that contains a primitive. This because methods like + Context(Part)>>reset have to be updated to skip the callPrimtiive bytecode otherwise." + "-1 for pre-increment in fetchNextBytecode" + ^thePC >= (objectMemory lastPointerOf: aMethod) + and: [thePC < (objectMemory numBytesOfBytes: aMethod)]!

1 0

VM Maker: VMMaker.oscog-eem.2181.mcz
by commits＠source.squeak.org 24 Mar '17

24 Mar '17

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.2181.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2181 Author: eem Time: 24 March 2017, 11:45:48.896507 am UUID: 7d1ebe24-d033-4fd1-8c67-52106d2c2a42 Ancestors: VMMaker.oscog-eem.2180 Make attemptToComputeTempNamesFor: robust enough to cope with AssemblerAbsentClassImports. Add the branchIf[Not]InstanceOf: bytecodes to the simulator's endPC calculation. =============== Diff against VMMaker.oscog-eem.2180 =============== Item was changed: ----- Method: Cogit class>>attemptToComputeTempNamesFor: (in category 'in-image compilation support') ----- attemptToComputeTempNamesFor: aCompiledMethod (aCompiledMethod respondsTo: #tempNames) ifTrue: [| blocks | blocks := aCompiledMethod embeddedBlockClosures. initializationOptions at: #tempNames put: (Dictionary withAll: {aCompiledMethod initialPC -> ([aCompiledMethod tempNames] on: MessageNotUnderstood do: [:ex| (self isSistaMessage: ex message unimplementedIn: Decompiler) ifTrue: [^self]. + (Smalltalk classNamed: #AssemblerAbsentClassImport) ifNotNil: + [:aaciClass| + (ex receiver isKindOf: aaciClass) ifTrue: + [^self]]. ex pass])}, (blocks ifEmpty: [#()] ifNotEmpty: [aCompiledMethod embeddedBlockClosures with: ((aCompiledMethod methodNode schematicTempNamesString allButFirst: (aCompiledMethod methodNode schematicTempNamesString indexOf: $[)) piecesCutWhere: [:a :b| b = $[]) collect: [:c :s| c startpc -> s substrings]]))]! Item was changed: ----- Method: StackInterpreterSimulator>>endPCOf: (in category 'compiled methods') ----- endPCOf: aMethod + "Determine the endPC of a method in the heap using interpretation that looks for returns and uses branches to skip intervening bytecodes." - "Determine the endPC of a method in the heap using interpretation that looks for returns." - - <var: #descriptor type: #'BytecodeDescriptor *'> | pc end farthestContinuation prim encoderClass inst is | (prim := self primitiveIndexOf: aMethod) > 0 ifTrue: [(self isQuickPrimitiveIndex: prim) ifTrue: [^(self startPCOfMethod: aMethod) - 1]]. encoderClass := self encoderClassForHeader: (objectMemory methodHeaderOf: aMethod). is := (InstructionStream on: (VMCompiledMethodProxy new for: aMethod coInterpreter: self objectMemory: objectMemory)). pc := farthestContinuation := self startPCOfMethod: aMethod. end := objectMemory numBytesOf: aMethod. is pc: pc + 1. [pc <= end] whileTrue: [inst := encoderClass interpretNextInstructionFor: MessageCatcher new in: is. inst selector caseOf: { [#pushClosureCopyNumCopiedValues:numArgs:blockSize:] -> [is pc: is pc + inst arguments last. farthestContinuation := farthestContinuation max: pc]. [#jump:] -> [farthestContinuation := farthestContinuation max: pc + inst arguments first]. [#jump:if:] -> [farthestContinuation := farthestContinuation max: pc + inst arguments first]. [#methodReturnConstant:] -> [pc >= farthestContinuation ifTrue: [end := pc]]. [#methodReturnReceiver] -> [pc >= farthestContinuation ifTrue: [end := pc]]. [#methodReturnTop] -> [pc >= farthestContinuation ifTrue: [end := pc]]. "This is for CompiledBlock/FullBlockClosure. Since the response to pushClosure... above skips over all block bytecoes, we will only see a blockReturnTop if it is at the top level, and so it must be a blockReturnTop in a CompiledBlock for a FullBlockClosure." + [#blockReturnTop] -> [pc >= farthestContinuation ifTrue: [end := pc]]. + [#branchIfInstanceOf:distance:] + -> [farthestContinuation := farthestContinuation max: pc + inst arguments last]. + [#branchIfNotInstanceOf:distance:] + -> [farthestContinuation := farthestContinuation max: pc + inst arguments last] } - [#blockReturnTop] -> [pc >= farthestContinuation ifTrue: [end := pc]] } otherwise: []. pc := is pc - 1]. ^end!

1 0

VM Maker: VMMaker.oscog-eem.2180.mcz
by commits＠source.squeak.org 24 Mar '17

24 Mar '17

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.2180.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2180 Author: eem Time: 24 March 2017, 9:26:28.687223 am UUID: 9866d19d-5f98-43c3-98ab-332b63416145 Ancestors: VMMaker.oscog-cb.2179 Stack depth checking for the full closure bytecode(s). =============== Diff against VMMaker.oscog-cb.2179 =============== Item was changed: ----- Method: StackDepthFinder>>pushFullClosure:numCopied: (in category 'instruction decoding') ----- + pushFullClosure: literal numCopied: numCopied - pushFullClosure: literal numCopied: numCopied. "Push Closure bytecode." numCopied = 0 ifTrue: [self push] ifFalse: [numCopied > 1 ifTrue: + [self drop: numCopied - 1]]! - [self drop: numCopied - 1]].! Item was added: + ----- Method: StackDepthFinder>>pushFullClosure:numCopied:receiverOnStack:ignoreOuterContext: (in category 'instruction decoding') ----- + pushFullClosure: literal numCopied: numCopied receiverOnStack: receiverOnStack ignoreOuterContext: ignoreOuterContext + "Push Closure bytecode." + | delta | + delta := 1 - (receiverOnStack ifTrue: [1] ifFalse: [0]) - numCopied. + delta ~= 0 ifTrue: [self drop: delta negated]!

1 0

VM Maker: VMMaker.oscog-cb.2179.mcz
by commits＠source.squeak.org 24 Mar '17

24 Mar '17

ClementBera uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-cb.2179.mcz ==================== Summary ==================== Name: VMMaker.oscog-cb.2179 Author: cb Time: 23 March 2017, 5:35:34.053468 pm UUID: 8a024d75-b1a4-4406-b88a-72c7a48692ad Ancestors: VMMaker.oscog-cb.2178 Fixed a minor bug in byteEqual Jitting. It looks stable now. =============== Diff against VMMaker.oscog-cb.2178 =============== Item was changed: ----- Method: SistaCogit>>genByteEqualsInlinePrimitive: (in category 'inline primitive generators') ----- genByteEqualsInlinePrimitive: prim "3021 Byte Object >> equals:length: The receiver and the arguments are both byte objects and have both the same size (length in bytes). The length argument is a smallinteger. Answers true if all fields are equal, false if not. Comparison is bulked to word comparison." "Overview: 1. The primitive is called like that: [byteObj1 equals: byteObj2 length: length]. In the worst case we use 5 registers including TempReg and we produce a loop bulk comparing words. 2. The common case is a comparison against a cst: [byteString = 'foo']. which produces in Scorch [byteString equals: 'foo' length: 3]. We try to generate fast code for this case with 3 heuristics: - specific fast code if len is a constant - unroll the loop if len < 2 * wordSize - compile-time reads if str1 or str2 is a constant and loop is unrolled. We use 3 registers including TempReg in the common case. We could use 1 less reg if the loop is unrolled, the instr is followed by a branch AND one operand is a constant, but this is complicated enough. 3. We ignore the case where all operands are constants (We assume Scorch simplifies it, it works but it is not optimised)" + | str1Reg str2Reg lenReg extraReg jmp jmp2 needjmpZeroSize needLoop unroll jmpZeroSize instr lenCst mask | - | str1Reg str2Reg lenReg extraReg jmp jmp2 adjust needjmpZeroSize needLoop unroll jmpZeroSize instr lenCst mask | <var: #jmp type: #'AbstractInstruction *'> <var: #instr type: #'AbstractInstruction *'> <var: #jmp2 type: #'AbstractInstruction *'> <var: #jmpZeroSize type: #'AbstractInstruction *'> "--- quick path for empty string---" "This path does not allocate registers and right shift on negative int later in the code. Normally this is resolved by Scorch but we keep it for correctness and consistency" self ssTop type = SSConstant ifTrue: [ lenCst := objectMemory integerValueOf: self ssTop constant. lenCst = 0 ifTrue: [ self ssPop: 3. self ssPushConstant: objectMemory trueObject. ^ 0 ] ]. "--- Allocating & loading registers --- " needLoop := (self ssTop type = SSConstant and: [ lenCst <= (objectMemory wordSize * 2) ]) not. unroll := needLoop not and: [lenCst > objectMemory wordSize ]. needLoop ifTrue: [ str1Reg := self allocateRegForStackEntryAt: 1 notConflictingWith: self emptyRegisterMask. str2Reg := self allocateRegForStackEntryAt: 2 notConflictingWith: (self registerMaskFor: str1Reg). + lenReg := self allocateRegForStackEntryAt: 0 notConflictingWith: (self registerMaskFor:str1Reg and: str2Reg). (self ssValue: 1) popToReg: str1Reg. (self ssValue: 2) popToReg: str2Reg. - lenReg := self allocateRegForStackEntryAt: 0 notConflictingWith: (self registerMaskFor:str1Reg and: str2Reg). extraReg := self allocateRegNotConflictingWith: (self registerMaskFor: str1Reg and: str2Reg and: lenReg)] ifFalse: [ mask := self emptyRegisterMask. (self ssValue: 1) type = SSConstant ifFalse: [ str1Reg := self allocateRegForStackEntryAt: 1 notConflictingWith: mask. (self ssValue: 1) popToReg: str1Reg. mask := mask bitOr: (self registerMaskFor: str1Reg) ]. (self ssValue: 2) type = SSConstant ifFalse: [ str2Reg := self allocateRegForStackEntryAt: 2 notConflictingWith: mask. (self ssValue: 2) popToReg: str2Reg. mask := mask bitOr: (self registerMaskFor: str2Reg) ]. extraReg := self allocateRegNotConflictingWith: mask]. - adjust := objectMemory baseHeaderSize >> objectMemory shiftForWord. "--- Loading LenReg (or statically resolving it) --- " "LenReg is loaded with (lenInBytes + objectMemory baseHeaderSize - 1 >> shiftForWord) LenReg is the index for the last word to compare with MoveXwr:r:R:. The loop iterates from LenReg to first word of ByteObj" self ssTop type = SSConstant ifTrue: "common case, str = 'foo'. We can precompute lenReg." [ lenCst := lenCst + objectMemory baseHeaderSize - 1 >> objectMemory shiftForWord. needLoop ifTrue: [self MoveCq: lenCst R: lenReg ]. needjmpZeroSize := false] ifFalse: "uncommon case, str = str2. lenReg in word computed at runtime." + [ self ssTop popToReg: lenReg. + objectRepresentation genConvertSmallIntegerToIntegerInReg: lenReg. - [ objectRepresentation genConvertSmallIntegerToIntegerInReg: lenReg. self CmpCq: 0 R: lenReg. jmpZeroSize := self JumpZero: 0. needjmpZeroSize := true. self AddCq: objectMemory baseHeaderSize - 1 R: lenReg. self ArithmeticShiftRightCq: objectMemory shiftForWord R: lenReg ]. "--- Comparing the strings --- " "LenReg has the index of the last word to read (unless no loop). We decrement it to adjust -1 (0 in 64 bits) while comparing" needLoop ifTrue: [instr := self MoveXwr: lenReg R: str1Reg R: extraReg. self MoveXwr: lenReg R: str2Reg R: TempReg. self CmpR: extraReg R: TempReg. jmp := self JumpNonZero: 0. "then string are not equal (jmp target)" self AddCq: -1 R: lenReg. + self CmpCq: (objectMemory baseHeaderSize >> objectMemory shiftForWord) - 1 R: lenReg. "first word of ByteObj, stop looping." - self CmpCq: adjust-1 R: lenReg. "first word of ByteObj, stop looping - in 64 bits adjust - 1 is 0, quicker." self JumpNonZero: instr] ifFalse: "Common case, only 1 or 2 word to check: no lenReg allocation, cst micro optimisations" [self genByteEqualsInlinePrimitiveCmp: str1Reg with: str2Reg scratch1: extraReg scratch2: TempReg field: 0. jmp := self JumpNonZero: 0. "then string are not equal (jmp target)" unroll ifTrue: "unrolling more than twice generate more instructions than the loop so we don't do it" [self genByteEqualsInlinePrimitiveCmp: str1Reg with: str2Reg scratch1: extraReg scratch2: TempReg field: 1. jmp2 := self JumpNonZero: 0. "then string are not equal (jmp target)"]]. needjmpZeroSize ifTrue: [ jmpZeroSize jmpTarget: self Label ]. "fall through, strings are equal" "--- Pushing the result or pipelining a branch --- " self ssPop: 3. self genByteEqualsInlinePrimitiveResult: jmp returnReg: extraReg. unroll ifTrue: [jmp2 jmpTarget: jmp getJmpTarget]. ^0!

1 0

VM Maker: BytecodeSets.spur-eem.75.mcz
by commits＠source.squeak.org 24 Mar '17

24 Mar '17

Eliot Miranda uploaded a new version of BytecodeSets to project VM Maker: http://source.squeak.org/VMMaker/BytecodeSets.spur-eem.75.mcz ==================== Summary ==================== Name: BytecodeSets.spur-eem.75 Author: eem Time: 23 March 2017, 5:19:46.501448 pm UUID: 44b5c3ef-e0ae-4e7d-8144-e8d60d5dd785 Ancestors: BytecodeSets.spur-eem.74 Use the ignoreOuterContext: form of the extended pushFullClosure: bytecode and nuke the outerContextNeeded: one. It's confusing. (And in doing so fix bugs in the gen method for the old variant). Do have the decoder issue the short pushFullClosure:numCopied: form if specified by the bytecode. =============== Diff against BytecodeSets.spur-eem.74 =============== Item was added: + ----- Method: BytecodeEncoder>>sizePushFullClosure:numCopied:receiverOnStack:ignoreOuterContext: (in category '*BytecodeSets-opcode sizing') ----- + sizePushFullClosure: compiledBlockLiteralIndex numCopied: numCopied receiverOnStack: rcvrOnStack ignoreOuterContext: ignoreOuterContext + ^self + sizeOpcodeSelector: #genPushFullClosure:numCopied:receiverOnStack:ignoreOuterContext: + withArguments: {compiledBlockLiteralIndex.numCopied. rcvrOnStack. ignoreOuterContext}! Item was changed: ----- Method: EncoderForSistaV1>>genPushFullClosure:numCopied: (in category 'extended bytecode generation') ----- genPushFullClosure: compiledBlockLiteralIndex numCopied: numCopied "By default the closure will have an outer context and the receiver will be fetched from the current context" + self genPushFullClosure: compiledBlockLiteralIndex numCopied: numCopied receiverOnStack: false ignoreOuterContext: false! - self genPushFullClosure: compiledBlockLiteralIndex numCopied: numCopied receiverOnStack: false outerContextNeeded: true! Item was added: + ----- Method: EncoderForSistaV1>>genPushFullClosure:numCopied:receiverOnStack:ignoreOuterContext: (in category 'extended bytecode generation') ----- + genPushFullClosure: compiledBlockLiteralIndex numCopied: numCopied receiverOnStack: receiverOnStack ignoreOuterContext: ignoreOuterContext + "* 249 11111001 xxxxxxxx siyyyyyy push Closure Compiled block literal index xxxxxxxx (+ Extend A * 256) numCopied yyyyyy receiverOnStack: s = 1 ignoreOuterContext: i = 1" + | extendedIndex | + (numCopied < 0 or: [numCopied > 64]) ifTrue: + [self outOfRangeError: 'num copied' index: numCopied range: 1 to: 64]. + (compiledBlockLiteralIndex < 0 or: [compiledBlockLiteralIndex > 32767]) ifTrue: + [^self outOfRangeError: 'index' index: compiledBlockLiteralIndex range: 0 to: 32767]. + (extendedIndex := compiledBlockLiteralIndex) > 255 ifTrue: + [self genUnsignedSingleExtendA: extendedIndex // 256. + extendedIndex := extendedIndex \\ 256]. + stream + nextPut: 249; + nextPut: extendedIndex; + nextPut: receiverOnStack asBit << 7 + (ignoreOuterContext asBit << 6) + numCopied! Item was removed: - ----- Method: EncoderForSistaV1>>genPushFullClosure:numCopied:receiverOnStack:outerContextNeeded: (in category 'extended bytecode generation') ----- - genPushFullClosure: compiledBlockLiteralIndex numCopied: numCopied receiverOnStack: receiverOnStack outerContextNeeded: outerContextNeeded - "* 249 11111001 xxxxxxxx siyyyyyy push Closure Compiled block literal index xxxxxxxx (+ Extend A * 256) numCopied yyyyyy receiverOnStack: s = 1 ignoreOuterContext: i = 1" - | extendedIndex | - (numCopied < 1 or: [numCopied > 64]) ifTrue: - [self outOfRangeError: 'num copied' index: numCopied range: 1 to: 64]. - (compiledBlockLiteralIndex < 0 or: [compiledBlockLiteralIndex > 32768]) ifTrue: - [^self outOfRangeError: 'index' index: compiledBlockLiteralIndex range: 0 to: 32768]. - (extendedIndex := compiledBlockLiteralIndex) > 255 ifTrue: - [self genUnsignedSingleExtendA: extendedIndex // 256. - extendedIndex := extendedIndex \\ 256]. - stream - nextPut: 249; - nextPut: extendedIndex; - nextPut: receiverOnStack asBit << 7 + (outerContextNeeded not asBit << 6) + numCopied! Item was added: + ----- Method: InstructionClient>>pushFullClosure:numCopied:receiverOnStack:ignoreOuterContext: (in category '*BytecodeSets-instruction decoding') ----- + pushFullClosure: compiledBlockLiteralIndex numCopied: numCopied receiverOnStack: rcvrOnStack ignoreOuterContext: ignoreOuterContext + "Creates and oushes a FullBlockClosure with variations." + ! Item was added: + ----- Method: InstructionPrinter>>pushFullClosure:numCopied:receiverOnStack:ignoreOuterContext: (in category '*BytecodeSets-instruction decoding') ----- + pushFullClosure: cb numCopied: num receiverOnStack: rcvrOnStack ignoreOuterContext: ignoreOuterContext + self print: 'pushFullClosure: ', cb printString, ' numCopied: ', num, ' receiverOnStack: ', rcvrOnStack printString, ' ignoreOuterContext: ', ignoreOuterContext printString! Item was changed: ----- Method: InstructionStream>>interpretNext3ByteSistaV1Instruction:for:extA:extB:startPC: (in category '*BytecodeSets-SistaV1-decoding') ----- interpretNext3ByteSistaV1Instruction: bytecode for: client extA: extA extB: extB startPC: startPC "Send to the argument, client, a message that specifies the next instruction. This method handles the three-byte codes. For a table of the bytecode set, see EncoderForSistaV1's class comment." | method byte2 byte3 literal | method := self method. byte2 := method at: pc. byte3 := method at: pc + 1. pc := pc + 2. "** 248 (2) 11111000 iiiiiiii mssjjjjj Call Primitive #iiiiiiii + (jjjjj * 256) m=1 means inlined primitive, no hard return after execution. ss defines the unsafe operation set used to encode the operations. (ss = 0 means sista unsafe operations, ss = 01 means lowcode operations, other numbers are not used) Lowcode inlined primitives may have extensions." bytecode = 248 ifTrue: [| primitiveSetSelector primitiveNumber | byte3 < 128 ifTrue: [ "Maybe this should be restricted to the 13 bit primitiveNumber too..." ^client callPrimitive: byte2 + (byte3 bitShift: 8)]. primitiveSetSelector := (byte3 bitShift: -5) bitAnd: 3. primitiveNumber := byte2 + ((byte3 bitAnd: 31) bitShift: 8). primitiveSetSelector = 0 ifTrue: "Sista inline primitives" [^client callInlinePrimitive: byte2 + (byte3 - 128 bitShift: 8)]. primitiveSetSelector = 1 ifTrue: "Lowcode inline primitives" [^client callLowcodeInlinePrimitive: byte2 + (byte3 - 128 bitShift: 8) extA: extA extB: extB]. "fall through to ^self unusedBytecode: client at: startPC below"]. "* 249 11111001 xxxxxxxx siyyyyyy push Closure Compiled block literal index xxxxxxxx (+ Extend A * 256) numCopied yyyyyy receiverOnStack: s = 1 ignoreOuterContext: i = 1" bytecode = 249 ifTrue: [literal := method literalAt: (extA bitShift: 8) + byte2 + 1. + (byte3 noMask: 16rC0) ifTrue: + [^client pushFullClosure: literal numCopied: byte3]. + ^client + pushFullClosure: literal + numCopied: (byte3 bitAnd: 16r3F) + receiverOnStack: (byte3 anyMask: 16r80) + ignoreOuterContext: (byte3 anyMask: 16r40)]. - ^client pushFullClosure: literal numCopied: (byte3 bitAnd: 16r3F)]. bytecode = 250 ifTrue: ["** 250 11111010 eeiiikkk jjjjjjjj Push Closure Num Copied iii (+ExtA//16*8) Num Args kkk (+ ExtA\\16*8) BlockSize jjjjjjjj (+ExtB*256). ee = num extensions" ^client pushClosureCopyNumCopiedValues: ((byte2 bitShift: -3) bitAnd: 7) + (extA // 16 bitShift: 3) numArgs: (byte2 bitAnd: 7) + (extA \\ 16 bitShift: 3) blockSize: byte3 + (extB bitShift: 8)]. bytecode = 251 ifTrue: [^client pushRemoteTemp: byte2 inVectorAt: byte3]. bytecode = 252 ifTrue: [^client storeIntoRemoteTemp: byte2 inVectorAt: byte3]. bytecode = 253 ifTrue: [^client popIntoRemoteTemp: byte2 inVectorAt: byte3]. "** 254 11111110 kkkkkkkk jjjjjjjj branch If Not Instance Of Behavior/Array Of Behavior kkkkkkkk (+ Extend A * 256, where Extend A >= 0) distance jjjjjjjj (+ Extend B * 256, where Extend B >= 0)" bytecode = 254 ifTrue: "The sign bit of extB inverts the operation. Would like to have extB < -128, but this is good enough for now." [literal := method literalAt: (extA bitShift: 8) + byte2 + 1. extB < 0 ifTrue: [^client branchIfInstanceOf: literal distance: (extB + 128 bitShift: 8) + byte3]. ^client branchIfNotInstanceOf: literal distance: (extB bitShift: 8) + byte3]. ^self unusedBytecode: client at: startPC!

1 0

VM Maker: VMMaker.oscog-cb.2178.mcz
by commits＠source.squeak.org 23 Mar '17

23 Mar '17

ClementBera uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-cb.2178.mcz ==================== Summary ==================== Name: VMMaker.oscog-cb.2178 Author: cb Time: 23 March 2017, 3:22:35.065129 pm UUID: a127d451-7d0d-440c-b339-aef5f18819bc Ancestors: VMMaker.oscog-cb.2177 Ooops. Forgot to remove debugging code. =============== Diff against VMMaker.oscog-cb.2177 =============== Item was changed: ----- Method: SistaCogit>>genByteEqualsInlinePrimitive: (in category 'inline primitive generators') ----- genByteEqualsInlinePrimitive: prim "3021 Byte Object >> equals:length: The receiver and the arguments are both byte objects and have both the same size (length in bytes). The length argument is a smallinteger. Answers true if all fields are equal, false if not. Comparison is bulked to word comparison." "Overview: 1. The primitive is called like that: [byteObj1 equals: byteObj2 length: length]. In the worst case we use 5 registers including TempReg and we produce a loop bulk comparing words. 2. The common case is a comparison against a cst: [byteString = 'foo']. which produces in Scorch [byteString equals: 'foo' length: 3]. We try to generate fast code for this case with 3 heuristics: - specific fast code if len is a constant - unroll the loop if len < 2 * wordSize - compile-time reads if str1 or str2 is a constant and loop is unrolled. We use 3 registers including TempReg in the common case. We could use 1 less reg if the loop is unrolled, the instr is followed by a branch AND one operand is a constant, but this is complicated enough. 3. We ignore the case where all operands are constants (We assume Scorch simplifies it, it works but it is not optimised)" | str1Reg str2Reg lenReg extraReg jmp jmp2 adjust needjmpZeroSize needLoop unroll jmpZeroSize instr lenCst mask | <var: #jmp type: #'AbstractInstruction *'> <var: #instr type: #'AbstractInstruction *'> <var: #jmp2 type: #'AbstractInstruction *'> <var: #jmpZeroSize type: #'AbstractInstruction *'> + - 1halt. "--- quick path for empty string---" "This path does not allocate registers and right shift on negative int later in the code. Normally this is resolved by Scorch but we keep it for correctness and consistency" self ssTop type = SSConstant ifTrue: [ lenCst := objectMemory integerValueOf: self ssTop constant. lenCst = 0 ifTrue: [ self ssPop: 3. self ssPushConstant: objectMemory trueObject. ^ 0 ] ]. + - lenCst := 7. "--- Allocating & loading registers --- " needLoop := (self ssTop type = SSConstant and: [ lenCst <= (objectMemory wordSize * 2) ]) not. unroll := needLoop not and: [lenCst > objectMemory wordSize ]. needLoop ifTrue: [ str1Reg := self allocateRegForStackEntryAt: 1 notConflictingWith: self emptyRegisterMask. str2Reg := self allocateRegForStackEntryAt: 2 notConflictingWith: (self registerMaskFor: str1Reg). (self ssValue: 1) popToReg: str1Reg. (self ssValue: 2) popToReg: str2Reg. lenReg := self allocateRegForStackEntryAt: 0 notConflictingWith: (self registerMaskFor:str1Reg and: str2Reg). extraReg := self allocateRegNotConflictingWith: (self registerMaskFor: str1Reg and: str2Reg and: lenReg)] ifFalse: [ mask := self emptyRegisterMask. (self ssValue: 1) type = SSConstant ifFalse: [ str1Reg := self allocateRegForStackEntryAt: 1 notConflictingWith: mask. (self ssValue: 1) popToReg: str1Reg. mask := mask bitOr: (self registerMaskFor: str1Reg) ]. (self ssValue: 2) type = SSConstant ifFalse: [ str2Reg := self allocateRegForStackEntryAt: 2 notConflictingWith: mask. (self ssValue: 2) popToReg: str2Reg. mask := mask bitOr: (self registerMaskFor: str2Reg) ]. extraReg := self allocateRegNotConflictingWith: mask]. adjust := objectMemory baseHeaderSize >> objectMemory shiftForWord. "--- Loading LenReg (or statically resolving it) --- " "LenReg is loaded with (lenInBytes + objectMemory baseHeaderSize - 1 >> shiftForWord) LenReg is the index for the last word to compare with MoveXwr:r:R:. The loop iterates from LenReg to first word of ByteObj" self ssTop type = SSConstant ifTrue: "common case, str = 'foo'. We can precompute lenReg." [ lenCst := lenCst + objectMemory baseHeaderSize - 1 >> objectMemory shiftForWord. needLoop ifTrue: [self MoveCq: lenCst R: lenReg ]. needjmpZeroSize := false] ifFalse: "uncommon case, str = str2. lenReg in word computed at runtime." [ objectRepresentation genConvertSmallIntegerToIntegerInReg: lenReg. self CmpCq: 0 R: lenReg. jmpZeroSize := self JumpZero: 0. needjmpZeroSize := true. self AddCq: objectMemory baseHeaderSize - 1 R: lenReg. self ArithmeticShiftRightCq: objectMemory shiftForWord R: lenReg ]. "--- Comparing the strings --- " "LenReg has the index of the last word to read (unless no loop). We decrement it to adjust -1 (0 in 64 bits) while comparing" needLoop ifTrue: [instr := self MoveXwr: lenReg R: str1Reg R: extraReg. self MoveXwr: lenReg R: str2Reg R: TempReg. self CmpR: extraReg R: TempReg. jmp := self JumpNonZero: 0. "then string are not equal (jmp target)" self AddCq: -1 R: lenReg. self CmpCq: adjust-1 R: lenReg. "first word of ByteObj, stop looping - in 64 bits adjust - 1 is 0, quicker." self JumpNonZero: instr] ifFalse: "Common case, only 1 or 2 word to check: no lenReg allocation, cst micro optimisations" [self genByteEqualsInlinePrimitiveCmp: str1Reg with: str2Reg scratch1: extraReg scratch2: TempReg field: 0. jmp := self JumpNonZero: 0. "then string are not equal (jmp target)" unroll ifTrue: "unrolling more than twice generate more instructions than the loop so we don't do it" [self genByteEqualsInlinePrimitiveCmp: str1Reg with: str2Reg scratch1: extraReg scratch2: TempReg field: 1. jmp2 := self JumpNonZero: 0. "then string are not equal (jmp target)"]]. needjmpZeroSize ifTrue: [ jmpZeroSize jmpTarget: self Label ]. "fall through, strings are equal" "--- Pushing the result or pipelining a branch --- " self ssPop: 3. self genByteEqualsInlinePrimitiveResult: jmp returnReg: extraReg. unroll ifTrue: [jmp2 jmpTarget: jmp getJmpTarget]. ^0!

1 0

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev March 2017