Vm-dev April 2015

vm-dev@lists.squeakfoundation.org

45 participants
264 discussions

NoDbgRegParms bug bites again--this time with CMake
by gettimothy 01 Apr '15

01 Apr '15

Just a heads up/progress report. The "fix" of putting a #define NoDbgRegParms "<--empty definition" in config.h.in for autotools does NOT translate to CMake because CMake builds its own config.h based on some tests/probing in a config.cmake file. This then creates its own config.h.in (cmake version) and then config.h When it does so, it does NOT respect the "empty definition" and just leave things alone, but "helps" by placing a /* #undef NoDbgRegParms */ in config.h There are 3 strategies I have in mind. 1. Have the user hand edit config.h (unacceptable, rejected) 2. Find a CMake workaround to generate the #define NoDbgRegParms 3. Take this as a clue about how autotools expects things and make the logic in Cog/src/vm/cogit.h Cog/src/vm/cointerp.h etc logic so that it works with #if !PRODUCTION && defined(__GNUC__) && !defined(NoDbgRegParms) # define NoDbgRegParms __attribute__ ((regparm (0))) #endif I will be focusing on approach 3 for a bit. "Maybe" its a. something about a comparison between the !defines and the !PRODUCTION b. a conflict created when autotools does its macro-expansion since the test is in multiple .h files. cheers. tty.

3 6

VM Maker: VMMaker.oscog-tpr.1143.mcz
by commits＠source.squeak.org 01 Apr '15

01 Apr '15

tim Rowledge uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-tpr.1143.mcz ==================== Summary ==================== Name: VMMaker.oscog-tpr.1143 Author: tpr Time: 31 March 2015, 9:11:29.066 pm UUID: 76e7613a-1b56-4e2b-9932-42df34c06102 Ancestors: VMMaker.oscog-eem.1142 More constant generation improvements in CogARMCompiler. Also some NSSendCacheSurrogate categorization changes that are there purely as artefacts of MC usage and don't actually represent any deliberate change at all. =============== Diff against VMMaker.oscog-eem.1142 =============== Item was removed: - ----- Method: CogARMCompiler>>ands:rn:rm:lsr: (in category 'ARM convenience instructions') ----- - ands: destReg rn: srcReg rm: addReg lsr: shft - "return an ANDS destReg, srcReg, addReg lsl #shft" - "important detail - a 0 shft requires setting the shift-type code to 0 to avoid potential instruction confusion" - shft = 0 - ifTrue:[^self type: 0 op: AndOpcode set: 1 rn: srcReg rd: destReg shifterOperand: addReg] - ifFalse:[^self type: 0 op: AndOpcode set: 1 rn: srcReg rd: destReg shifterOperand: ((shft <<7 bitOr: 32) bitOr: addReg)]! Item was changed: ----- Method: CogARMCompiler>>concretizeAndCqR (in category 'generate machine code - concretize') ----- concretizeAndCqR "Will get inlined into concretizeAt: switch." "AND is very important since it's used to mask all sorts of flags in the jit. We take special care to try to find fast ways to make the masks" <inline: true> |val| val := operands at: 0. self rotateable8bitImmediate: val ifTrue: [ :rot :immediate | | reg | reg := self concreteRegister: (operands at: 1). self machineCodeAt: 0 put: (self ands: reg rn: reg imm: immediate ror: rot). ^machineCodeSize := 4] ifFalse: [ "see if the constant bit-inverted makes a quick value and if so BIC it instead If the value is -ve, we 2s complement it instead" |invVal| val <0 ifTrue:[invVal := -1 - val] ifFalse:[invVal := val bitInvert32]. self rotateable8bitImmediate: invVal ifTrue: [ :rot :immediate | |reg| reg := self concreteRegister: (operands at: 1). self machineCodeAt: 0 put: (self bics: reg rn: reg imm: immediate ror: rot). ^machineCodeSize := 4] ifFalse: ["let's try to see if the constant can be made from a simple shift of 0xFFFFFFFF" |hb reg| reg := self concreteRegister: (operands at: 1). hb := (operands at: 0) highBit. 1 << hb = (val +1) ifTrue: [ "MVN temp reg, 0, making 0xffffffff" self machineCodeAt: 0 put:(self mvn: ConcreteIPReg imm: 0 ror: 0). "Then AND reg, temp reg, lsr #(32-hb)" + self machineCodeAt: 4 put:(self dataOpType: AndOpcode rd: reg rn: reg rm: ConcreteIPReg lsr: (32-hb )). - self machineCodeAt: 4 put:(self ands: reg rn: reg rm: ConcreteIPReg lsr: (32-hb )). ^machineCodeSize :=8] ifFalse: [^self concretizeDataOperationCwR: AndOpcode]]]! Item was changed: ----- Method: CogARMCompiler>>concretizeAndCqRR (in category 'generate machine code - concretize') ----- concretizeAndCqRR "Will get inlined into concretizeAt: switch." "AND is very important since it's used to mask all sorts of flags in the jit. We take special care to try to find fast ways to make the masks" <inline: true> | val srcReg dstReg | val := operands at: 0. srcReg := self concreteRegister: (operands at: 1). dstReg := self concreteRegister: (operands at: 2). self rotateable8bitImmediate: val ifTrue: [ :rot :immediate | self machineCodeAt: 0 put: (self ands: dstReg rn: srcReg imm: immediate ror: rot). ^machineCodeSize := 4] ifFalse: ["see if the constant bit-inverted makes a quick value and if so BIC it instead If the value is -ve, we 2s complement it instead" |invVal| invVal := val < 0 ifTrue:[-1 - val] ifFalse:[val bitInvert32]. self rotateable8bitImmediate: invVal ifTrue: [ :rot :immediate | self machineCodeAt: 0 put: (self bics: dstReg rn: srcReg imm: immediate ror: rot). ^machineCodeSize := 4] ifFalse: "let's try to see if the constant can be made from a simple shift of 0xFFFFFFFF" [| hb | hb := (operands at: 0) highBit. 1 << hb = (val +1) ifTrue: "MVN temp reg, 0, making 0xffffffff" [self machineCodeAt: 0 put:(self mvn: ConcreteIPReg imm: 0 ror: 0). "Then AND reg, temp reg, lsr #(32-hb)" + self machineCodeAt: 4 put: (self dataOpType: AndOpcode rd: dstReg rn: srcReg rm: ConcreteIPReg lsr: 32 - hb). - self machineCodeAt: 4 put: (self ands: dstReg rn: srcReg rm: ConcreteIPReg lsr: 32 - hb). ^machineCodeSize := 8] ifFalse: [^self concretizeDataOperationCwR: AndOpcode]]]! Item was changed: ----- Method: CogARMCompiler>>concretizeDataOperationCqR: (in category 'generate machine code - concretize') ----- concretizeDataOperationCqR: armOpcode "Will get inlined into concretizeAt: switch." "4 == Add, 2 == Sub, Xor == 1, And == 0, Or == 12, Bic == 14" <inline: true> + |val rd rn | + val := operands at: 0. + rn := self concreteRegister: (operands at: 1). + rd := opcode = CmpOpcode ifTrue: [0] ifFalse:[rn]. + + self rotateable8bitImmediate: val + ifTrue: [:rot :immediate | - self - rotateable8bitImmediate: (operands at: 0) - ifTrue: [:rot :immediate | | rd rn | - rn := self concreteRegister: (operands at: 1). - rd := opcode = CmpOpcode ifTrue: [0] ifFalse:[rn]. self machineCodeAt: 0 put: (self type: 1 op: armOpcode set: 1 rn: rn rd: rd shifterOperand: ((rot>>1)"in this usage we have to halve the rot value" << 8 bitOr: immediate)). ^machineCodeSize := 4] + ifFalse: ["let's try to see if the constant can be made from a simple shift of 0xFFFFFFFF" + val > 0 ifTrue: [ + |hb | + hb := val highBit. + 1 << hb = (val +1) + ifTrue: [ "MVN temp, #0, making 0xffffffff" + self machineCodeAt: 0 put:(self mvn: ConcreteIPReg imm: 0 ror: 0). + "Then armOpcode reg, temp reg, lsr #(32-hb)" + self machineCodeAt: 4 put:(self dataOpType: armOpcode rd: rd rn: rn rm: ConcreteIPReg lsr: (32-hb)). + ^machineCodeSize :=8]]. + ^self concretizeDataOperationCwR: armOpcode]. - ifFalse: [^self concretizeDataOperationCwR: armOpcode]. ! Item was added: + ----- Method: CogARMCompiler>>dataOpType:rd:rn:rm:lsr: (in category 'ARM convenience instructions') ----- + dataOpType: opcode rd: destReg rn: srcReg rm: addReg lsr: shft + "return an {opcode} destReg, srcReg, addReg lsl #shft" + "important detail - a 0 shft requires setting the shift-type code to 0 to avoid potential instruction confusion" + shft = 0 + ifTrue:[^self type: 0 op: opcode set: 1 rn: srcReg rd: destReg shifterOperand: addReg] + ifFalse:[^self type: 0 op: opcode set: 1 rn: srcReg rd: destReg shifterOperand: ((shft <<7 bitOr: 32) bitOr: addReg)]! Item was changed: SharedPool subclass: #CogRTLOpcodes instanceVariableNames: '' + classVariableNames: 'AddCqR AddCwR AddRR AddRdRd AlignmentNops AndCqR AndCqRR AndCwR AndRR Arg0Reg Arg1Reg ArithmeticShiftRightCqR ArithmeticShiftRightRR Call ClassReg CmpCqR CmpCwR CmpRR CmpRdRd ConvertRRd DPFPReg0 DPFPReg1 DPFPReg2 DPFPReg3 DPFPReg4 DPFPReg5 DPFPReg6 DPFPReg7 DivRdRd FPReg Fill16 Fill32 Fill8 FillBytesFrom FillFromWord FirstJump FirstShortJump GPRegMax GPRegMin Jump JumpAbove JumpAboveOrEqual JumpBelow JumpBelowOrEqual JumpCarry JumpFPEqual JumpFPGreater JumpFPGreaterOrEqual JumpFPLess JumpFPLessOrEqual JumpFPNotEqual JumpFPOrdered JumpFPUnordered JumpGreater JumpGreaterOrEqual JumpLess JumpLessOrEqual JumpLong JumpLongNonZero JumpLongZero JumpNegative JumpNoCarry JumpNoOverflow JumpNonNegative JumpNonZero JumpOverflow JumpR JumpZero Label LastJump LastRTLCode LinkReg LoadEffectiveAddressMwrR LoadEffectiveAddressXowrR LogicalShiftLeftCqR LogicalShiftLeftRR LogicalShiftRightCqR LogicalShiftRightRR MoveAbR MoveAwR MoveC32R MoveC64R MoveCqR MoveCwR MoveM16rR MoveM32rR MoveM64rRd MoveMbrR MoveMwrR MoveRAb MoveRAw MoveRM16r MoveRM32r MoveRMbr MoveRMwr MoveRR MoveRX16rR MoveRX32rR MoveRXbrR MoveRXowr MoveRXwrR MoveRdM64r MoveRdRd MoveX16rRR MoveX32rRR MoveXbrRR MoveXowrR MoveXwrRR MulCqR MulCwR MulRR MulRdRd NegateR Nop OrCqR OrCwR OrRR PCReg PopR PrefetchAw PushCw PushR RISCTempReg ReceiverResultReg RetN SPReg SendNumArgsReg SqrtRd SubCqR SubCwR SubRR SubRdRd TempReg TstCqR UnusedARMRegister VarBaseReg XorCqR XorCwR XorRR' - classVariableNames: 'AddCqR AddCwR AddRR AddRdRd AlignmentNops AndCqR AndCqRR AndCwR AndRR Arg0Reg Arg1Reg ArithmeticShiftRightCqR ArithmeticShiftRightRR Call ClassReg CmpCqR CmpCwR CmpRR CmpRdRd ConvertRRd DPFPReg0 DPFPReg1 DPFPReg2 DPFPReg3 DPFPReg4 DPFPReg5 DPFPReg6 DPFPReg7 DivRdRd FPReg Fill16 Fill32 Fill8 FillBytesFrom FillFromWord FirstJump FirstShortJump GPRegMax GPRegMin Jump JumpAbove JumpAboveOrEqual JumpBelow JumpBelowOrEqual JumpCarry JumpFPEqual JumpFPGreater JumpFPGreaterOrEqual JumpFPLess JumpFPLessOrEqual JumpFPNotEqual JumpFPOrdered JumpFPUnordered JumpGreater JumpGreaterOrEqual JumpLess JumpLessOrEqual JumpLong JumpLongNonZero JumpLongZero JumpNegative JumpNoCarry JumpNoOverflow JumpNonNegative JumpNonZero JumpOverflow JumpR JumpZero Label LastJump LastRTLCode LinkReg LoadEffectiveAddressMwrR LoadEffectiveAddressXowrR LogicalShiftLeftCqR LogicalShiftLeftRR LogicalShiftRightCqR LogicalShiftRightRR MoveAbR MoveAwR MoveC32R MoveC64R MoveCqR MoveCwR MoveM16rR MoveM32rR MoveM64rRd MoveMbrR MoveMwrR MoveRAb MoveRAw MoveRM16r MoveRM32r MoveRMbr MoveRMwr MoveRR MoveRX16rR MoveRX32rR MoveRXbrR MoveRXowr MoveRXwrR MoveRdM64r MoveRdRd MoveX16rRR MoveX32rRR MoveXbrRR MoveXowrR MoveXwrRR MulCqR MulCwR MulRR MulRdRd NegateR Nop OrCqR OrCwR OrRR PCReg PopR PrefetchAw PushCw PushR RISCTempReg ReceiverResultReg RetN SPReg SendNumArgsReg SqrtRd SubCqR SubCwR SubRR SubRdRd TempReg TstCqR VarBaseReg XorCqR XorCwR XorRR' poolDictionaries: '' category: 'VMMaker-JIT'! !CogRTLOpcodes commentStamp: '<historical>' prior: 0! I am a pool for the Register-Transfer-Language to which Cog compiles. I define unique integer values for all RTL opcodes and abstract registers. See CogAbstractInstruction for instances of instructions with the opcodes that I define.! Item was changed: + ----- Method: NSSendCacheSurrogate32 class>>alignedByteSize (in category 'accessing') ----- - ----- Method: NSSendCacheSurrogate32 class>>alignedByteSize (in category 'as yet unclassified') ----- alignedByteSize ^20! Item was changed: + ----- Method: NSSendCacheSurrogate32>>classTag (in category 'accessing') ----- - ----- Method: NSSendCacheSurrogate32>>classTag (in category 'as yet unclassified') ----- classTag ^memory unsignedLongAt: address + 9! Item was changed: + ----- Method: NSSendCacheSurrogate32>>enclosingObject (in category 'accessing') ----- - ----- Method: NSSendCacheSurrogate32>>enclosingObject (in category 'as yet unclassified') ----- enclosingObject ^memory unsignedLongAt: address + 13! Item was changed: ----- Method: NSSendCacheSurrogate32>>enclosingObject: (in category 'accessing') ----- enclosingObject: aValue ^memory unsignedLongAt: address + 13 put: aValue! Item was changed: + ----- Method: NSSendCacheSurrogate32>>numArgs (in category 'accessing') ----- - ----- Method: NSSendCacheSurrogate32>>numArgs (in category 'as yet unclassified') ----- numArgs ^memory unsignedLongAt: address + 5! Item was changed: + ----- Method: NSSendCacheSurrogate32>>selector (in category 'accessing') ----- - ----- Method: NSSendCacheSurrogate32>>selector (in category 'as yet unclassified') ----- selector ^memory unsignedLongAt: address + 1! Item was changed: + ----- Method: NSSendCacheSurrogate32>>target (in category 'accessing') ----- - ----- Method: NSSendCacheSurrogate32>>target (in category 'as yet unclassified') ----- target ^memory unsignedLongAt: address + 17! Item was changed: + ----- Method: NSSendCacheSurrogate64 class>>alignedByteSize (in category 'accessing') ----- - ----- Method: NSSendCacheSurrogate64 class>>alignedByteSize (in category 'as yet unclassified') ----- alignedByteSize ^40!

1 0

VM Maker: VMMaker.oscog-eem.1142.mcz
by commits＠source.squeak.org 01 Apr '15

01 Apr '15

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.1142.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.1142 Author: eem Time: 31 March 2015, 7:03:04.226 pm UUID: 518e91ab-2513-4368-acb6-830554983a2b Ancestors: VMMaker.oscog-eem.1141 Make sure the stack is scanned after a pin operation that involves forwarding. Make sure that the unforwarding stack scan post-become (or post-pin) scans all of the stack contents in the CoInterpreter, sicne super sends have no read barrier. Implement the SistaV1 directed super send bytecode in the StackInterpreter. Fix BitBlt simulation: - BitBltSimulator>>halftoneAt: does not work in Cog; nuke it. - BitBltSimulator>>initialiseModule must answer a boolean. - tryLoadNewPlugin:pluginEntries: needed to choose the plugin classes more carefully. =============== Diff against VMMaker.oscog-eem.1141 =============== Item was removed: - ----- Method: BitBltSimulator>>halftoneAt: (in category 'simulation') ----- - halftoneAt: idx - - ^ (halftoneBase + (idx \\ halftoneHeight * 4)) long32At: 0! Item was changed: ----- Method: BitBltSimulator>>initialiseModule (in category 'initialize-release') ----- initialiseModule self class isInitialised ifFalse: [| ivars | ivars := #(opTable maskTable warpBitShiftTable ditherMatrix4x4 ditherThresholds16 ditherValues16 dither8Lookup). super initialiseModule. ivars do: [:symbol | self class instVarNamed: symbol put: (self instVarNamed: symbol)]. self class setInitialised]. opTable := self class opTable. maskTable := self class maskTable. warpBitShiftTable := self class warpBitShiftTable. ditherMatrix4x4 := self class ditherMatrix4x4. ditherThresholds16 := self class ditherThresholds16. ditherValues16 := self class ditherValues16. dither8Lookup := self class dither8Lookup. + ^true! - ! Item was changed: ----- Method: CoInterpreter>>followForwardingPointersInStackZone: (in category 'object memory support') ----- followForwardingPointersInStackZone: theBecomeEffectsFlags "Spur's become: is lazy, turning the becommed object into a forwarding object to the other. The read-barrier is minimised by arranging that forwarding pointers will fail a method cache probe, since notionally objects' internals are accessed only via sending messages to them, the exception is primitives that access the internals of the non-receiver argument(s). To avoid a read barrier on inst var fetch we scan the receivers in the stack zone and follow + any forwarded ones. This is way cheaper than scanning all of memory as in the old become. + + Override to handle machine code frames, and to handle the lack of an explicit read barrier on super sends. + With most super send implementations (not Newspeak's absent super bytecodes) self, the receiver of the + super send, is pushed before any arguments. So if self is becommed during argument marshalling, e.g. + super doSomethingWith: (self become: self somethingElse) + then a stale forwarded reference to self could be left on the stack. In the StackInterpreter we deal with this + with an explicit read barrier on supersend. In the CoInterpreter we deal with it by following all non-argument + stack contents." - any forwarded ones. This is way cheaper than scanning all of memory as in the old become." | theIPPtr | <inline: false> <var: #thePage type: #'StackPage *'> <var: #theSP type: #'char *'> <var: #theFP type: #'char *'> <var: #callerFP type: #'char *'> <var: #theIPPtr type: #usqInt> + self externalWriteBackHeadFramePointers. + (theBecomeEffectsFlags anyMask: BecameCompiledMethodFlag) ifTrue: [(objectMemory isForwarded: method) ifTrue: [theIPPtr := instructionPointer - method. method := objectMemory followForwarded: method. instructionPointer := method + theIPPtr]. (objectMemory isOopForwarded: newMethod) ifTrue: [newMethod := objectMemory followForwarded: newMethod]]. self assert: stackPage ~= 0. 0 to: numStackPages - 1 do: [:i| | thePage theSP theFP callerFP oop offset | thePage := stackPages stackPageAt: i. thePage isFree ifFalse: [theSP := thePage headSP. theFP := thePage headFP. "Skip the instruction pointer on top of stack of inactive pages." thePage = stackPage ifTrue: [theIPPtr := 0] ifFalse: [theIPPtr := theSP asUnsignedInteger. theSP := theSP + objectMemory wordSize]. [self assert: (thePage addressIsInPage: theFP). self assert: (theIPPtr = 0 or: [thePage addressIsInPage: theIPPtr asVoidPointer]). offset := self frameStackedReceiverOffset: theFP. + [theSP <= offset] whileTrue: + [oop := stackPages longAt: theSP. + (objectMemory isOopForwarded: oop) ifTrue: + [stackPages longAt: theSP put: (objectMemory followForwarded: oop)]. + theSP := theSP + objectMemory wordSize]. - oop := stackPages longAt: theFP + offset. - (objectMemory isOopForwarded: oop) ifTrue: - [stackPages - longAt: theFP + offset - put: (objectMemory followForwarded: oop)]. ((self frameHasContext: theFP) and: [(objectMemory isForwarded: (self frameContext: theFP))]) ifTrue: [stackPages longAt: theFP + FoxThisContext put: (objectMemory followForwarded: (self frameContext: theFP))]. (self isMachineCodeFrame: theFP) ifTrue: [oop := stackPages longAt: theFP + FoxMFReceiver. (objectMemory isOopForwarded: oop) ifTrue: [stackPages longAt: theFP + FoxMFReceiver put: (objectMemory followForwarded: oop)]. oop := (self mframeHomeMethod: theFP) methodObject. self assert: (objectMemory isForwarded: oop) not] ifFalse: [oop := stackPages longAt: theFP + FoxIFReceiver. (objectMemory isOopForwarded: oop) ifTrue: [stackPages longAt: theFP + FoxIFReceiver put: (objectMemory followForwarded: oop)]. oop := self iframeMethod: theFP. (objectMemory isForwarded: oop) ifTrue: [| newOop delta | newOop := objectMemory followForwarded: oop. delta := newOop - oop. (theIPPtr ~= 0 and: [(stackPages longAt: theIPPtr) > oop]) ifTrue: [stackPages longAt: theIPPtr put: (stackPages longAt: theIPPtr) + delta]. stackPages longAt: theFP + FoxIFSavedIP put: (stackPages longAt: theFP + FoxIFSavedIP) + delta. stackPages longAt: theFP + FoxMethod put: (oop := newOop)]]. (callerFP := self frameCallerFP: theFP) ~= 0] whileTrue: [theIPPtr := (theFP + FoxCallerSavedIP) asUnsignedInteger. + theSP := theIPPtr + objectMemory wordSize. theFP := callerFP]. "And finally follow the saved context and the caller context." theSP := thePage baseAddress - objectMemory wordSize. [theSP <= thePage baseAddress] whileTrue: [oop := stackPages longAt: theSP. (objectMemory isForwarded: oop) ifTrue: [stackPages longAt: theSP put: (objectMemory followForwarded: oop)]. theSP := theSP + objectMemory wordSize]]]! Item was changed: ----- Method: SpurMemoryManager>>pinObject: (in category 'primitive support') ----- pinObject: objOop "Attempt to pin objOop, which must not be immediate. If the attempt succeeds answer objOop's (possibly moved) oop. If the attept fails, which can only occur if there is no memory, answer 0." <inline: false> | oldClone seg | <var: #seg type: #'SpurSegmentInfo *'> self assert: (self isNonImmediate: objOop). self flag: 'policy decision here. if already old, do we clone in a segment containing pinned objects or merely pin?'. "We choose to clone to keep pinned objects together to reduce fragmentation, assuming that pinning is rare and that fragmentation is a bad thing." (self isOldObject: objOop) ifTrue: [seg := segmentManager segmentContainingObj: objOop. seg containsPinned ifTrue: [self setIsPinnedOf: objOop to: true. ^objOop]. segmentManager someSegmentContainsPinned ifFalse: [self setIsPinnedOf: objOop to: true. seg containsPinned: true. ^objOop]]. oldClone := self cloneInOldSpaceForPinning: objOop. oldClone ~= 0 ifTrue: + [becomeEffectsFlags := self becomeEffectFlagsFor: objOop. + self setIsPinnedOf: oldClone to: true. + self forward: objOop to: oldClone. + coInterpreter postBecomeAction: becomeEffectsFlags. + self postBecomeScanClassTable: becomeEffectsFlags. + becomeEffectsFlags := 0]. - [self setIsPinnedOf: oldClone to: true. - self forward: objOop to: oldClone]. ^oldClone! Item was added: + ----- Method: StackInterpreter>>directedSuperclassSend (in category 'send bytecodes') ----- + directedSuperclassSend + "Send a message to self, starting lookup with the superclass of the class on top of stack." + "Assume: messageSelector and argumentCount have been set, and that + the receiver and arguments have been pushed onto the stack," + "Note: This method is inlined into the interpreter dispatch loop." + <sharedCodeInCase: #singleExtendedSuperBytecode> + <option: #SistaVM> + | class superclass | + class := self internalPopStack. + (objectMemory isForwarded: class) ifTrue: + [class := objectMemory followForwarded: class]. + superclass := self superclassOf: class. + objectMemory ensureBehaviorHash: superclass. + lkupClassTag := objectMemory classTagForClass: superclass. + "To maintain the invariant that all receivers are unforwarded we need an explicit + read barrier in the super send cases. Even though we always follow receivers + on become e.g. super doSomethingWith: (self become: other) forwards the receiver + self pushed on the stack." + self ensureReceiverUnforwarded. + self assert: lkupClassTag ~= objectMemory nilObject. + self commonSend! Item was changed: ----- Method: StackInterpreter>>extSendSuperBytecode (in category 'send bytecodes') ----- extSendSuperBytecode + "239 11101111 i i i i i j j j + ExtendB < 64 + ifTrue: [Send To Superclass Literal Selector #iiiii (+ Extend A * 32) with jjj (+ Extend B * 8) Arguments] + ifFalse: [Send To Superclass of Stacked Class Literal Selector #iiiii (+ Extend A * 32) with jjj (+ (Extend B bitAnd: 63) * 8) Arguments]" - "239 11101111 i i i i i j j j Send To Superclass Literal Selector #iiiii (+ Extend A * 32) with jjj (+ Extend B * 8) Arguments" | byte | byte := self fetchByte. messageSelector := self literal: (byte >> 3) + (extA << 5). extA := 0. + SistaVM ifTrue: + [extB >= 64 ifTrue: + [argumentCount := (byte bitAnd: 7) + (extB << 3). + extB := 0. + ^self directedSuperclassSend]]. argumentCount := (byte bitAnd: 7) + (extB << 3). extB := 0. + self superclassSend! - self superclassSend.! Item was changed: ----- Method: StackInterpreter>>followForwardingPointersInStackZone: (in category 'object memory support') ----- followForwardingPointersInStackZone: theBecomeEffectsFlags "Spur's become: is lazy, turning the becommed object into a forwarding object to the other. The read-barrier is minimised by arranging that forwarding pointers will fail a method cache probe, since notionally objects' internals are accessed only via sending messages to them, the exception is primitives that access the internals of the non-receiver argument(s). To avoid a read barrier on bytecode, literal and inst var fetch we scan the receivers and methods in the stack zone and follow any forwarded ones. This is of course way cheaper than scanning all of memory as in the old become." | theIPPtr | <inline: false> <var: #thePage type: #'StackPage *'> <var: #theSP type: #'char *'> <var: #theFP type: #'char *'> <var: #callerFP type: #'char *'> <var: #theIPPtr type: #usqInt> + self externalWriteBackHeadFramePointers. + (theBecomeEffectsFlags anyMask: BecameCompiledMethodFlag) ifTrue: [(objectMemory isForwarded: method) ifTrue: [theIPPtr := instructionPointer - method. method := objectMemory followForwarded: method. instructionPointer := method + theIPPtr]. (objectMemory isOopForwarded: newMethod) ifTrue: [newMethod := objectMemory followForwarded: newMethod]]. self assert: stackPage ~= 0. 0 to: numStackPages - 1 do: [:i| | thePage theSP theFP callerFP theIP oop | thePage := stackPages stackPageAt: i. thePage isFree ifFalse: [theSP := thePage headSP. theFP := thePage headFP. "Skip the instruction pointer on top of stack of inactive pages." thePage = stackPage ifTrue: [theIPPtr := 0] ifFalse: [theIPPtr := theSP asInteger. theSP := theSP + objectMemory wordSize]. [self assert: (thePage addressIsInPage: theFP). self assert: (theIPPtr = 0 or: [thePage addressIsInPage: theIPPtr asVoidPointer]). oop := stackPages longAt: theFP + FoxReceiver. (objectMemory isOopForwarded: oop) ifTrue: [stackPages longAt: theFP + FoxReceiver put: (objectMemory followForwarded: oop)]. theIP := (theFP + (self frameStackedReceiverOffset: theFP)) asInteger. "reuse theIP; its just an offset here" oop := stackPages longAt: theIP. (objectMemory isOopForwarded: oop) ifTrue: [stackPages longAt: theIP put: (objectMemory followForwarded: oop)]. ((self frameHasContext: theFP) and: [(objectMemory isForwarded: (self frameContext: theFP))]) ifTrue: [stackPages longAt: theFP + FoxThisContext put: (objectMemory followForwarded: (self frameContext: theFP))]. oop := self frameMethod: theFP. (objectMemory isForwarded: oop) ifTrue: [| newOop delta | newOop := objectMemory followForwarded: oop. theIPPtr ~= 0 ifTrue: [self assert: (stackPages longAt: theIPPtr) > (self frameMethod: theFP). delta := newOop - oop. stackPages longAt: theIPPtr put: (stackPages longAt: theIPPtr) + delta]. stackPages longAt: theFP + FoxMethod put: (oop := newOop)]. (callerFP := self frameCallerFP: theFP) ~= 0] whileTrue: [theIPPtr := (theFP + FoxCallerSavedIP) asInteger. theFP := callerFP]. "And finally follow the caller context." self assert: theFP = thePage baseFP. oop := self frameCallerContext: theFP. (objectMemory isForwarded: oop) ifTrue: [self frameCallerContext: theFP put: (objectMemory followForwarded: oop)]]]! Item was changed: ----- Method: StackInterpreter>>tryLoadNewPlugin:pluginEntries: (in category 'primitive support') ----- tryLoadNewPlugin: pluginString pluginEntries: pluginEntries "Load the plugin and if on Spur, populate pluginEntries with the prmitives in the plugin." <doNotGenerate> | plugin realPluginClass plugins simulatorClasses | self transcript cr; show: 'Looking for module ', pluginString. "Defeat loading of the FloatArrayPlugin & Matrix2x3Plugin since complications with 32-bit float support prevent simulation. If you feel up to tackling this start by implementing cCoerce: value to: cType ^cType = 'float' ifTrue: [value asIEEE32BitWord] ifFalse: [value] in FloatArrayPlugin & Matrix2x3Plugin and then address the issues in the BalloonEnginePlugin. See http://forum.world.st/Simulating-the-BalloonEnginePlugin-FloatArrayPlugin-a…" (#('FloatArrayPlugin' 'Matrix2x3Plugin') includes: pluginString) ifTrue: [self transcript show: ' ... defeated'. ^nil]. pluginString isEmpty ifTrue: [plugin := self] ifFalse: [plugins := InterpreterPlugin allSubclasses select: [:psc| psc moduleName asString = pluginString asString]. simulatorClasses := (plugins select: [:psc| psc simulatorClass notNil] thenCollect: [:psc| psc simulatorClass]) asSet. simulatorClasses isEmpty ifTrue: [self transcript show: ' ... not found'. ^nil]. simulatorClasses size > 1 ifTrue: [^self error: 'This won''t work...']. + (plugins copyWithoutAll: simulatorClasses) notEmpty ifTrue: + [plugins := plugins copyWithoutAll: simulatorClasses]. plugins size > 1 ifTrue: [self transcript show: '...multiple plugin classes; choosing ', plugins last name]. realPluginClass := plugins last. "hopefully lowest in the hierarchy..." plugin := simulatorClasses anyOne newFor: realPluginClass. plugin setInterpreter: objectMemory. "Ignore return value from setInterpreter" (plugin respondsTo: #initialiseModule) ifTrue: [plugin initialiseModule ifFalse: [self transcript show: ' ... initialiser failed'. ^nil]]]. "module initialiser failed" self transcript show: ' ... loaded'. objectMemory hasSpurMemoryManagerAPI ifTrue: [| realPlugin cg | self transcript show: '...computing accessor depths'. plugin class isPluginClass ifTrue: [realPlugin := (plugin isSmartSyntaxPluginSimulator ifTrue: [realPluginClass] ifFalse: [plugin class]) withAllSuperclasses detect: [:class| class shouldBeTranslated]. cg := realPlugin buildCodeGeneratorUpTo: realPlugin] ifFalse: [cg := self codeGeneratorToComputeAccessorDepth. primitiveTable withIndexDo: [:prim :index| | depth | prim isSymbol ifTrue: [depth := cg accessorDepthForSelector: prim. self assert: (depth isInteger or: [depth isNil and: [(plugin class whichClassIncludesSelector: prim) isNil]]). primitiveAccessorDepthTable at: index - 1 put: depth]]]. cg exportedPrimitiveNames do: [:primName| | fnSymbol | fnSymbol := primName asSymbol. pluginEntries addLast: {plugin. fnSymbol. [plugin perform: fnSymbol. self]. cg accessorDepthForSelector: fnSymbol}]. self transcript show: '...done']. ^pluginString asString -> plugin!

1 0

VM Maker: BytecodeSets.spur-eem.20.mcz
by commits＠source.squeak.org 01 Apr '15

01 Apr '15

Eliot Miranda uploaded a new version of BytecodeSets to project VM Maker: http://source.squeak.org/VMMaker/BytecodeSets.spur-eem.20.mcz ==================== Summary ==================== Name: BytecodeSets.spur-eem.20 Author: eem Time: 31 March 2015, 5:20:37.942 pm UUID: 99c1c437-29d1-4907-ae45-67885077ea1b Ancestors: BytecodeSets.spur-eem.19 Redefine the SistaV1 super send bytecode to encompass a directed super send for inlining super sends in optimized methods where the super send may be form a method inlined from some superclass and hence /not/ the superclass of the optimized method's mthod class. =============== Diff against BytecodeSets.spur-eem.19 =============== Item was changed: BytecodeEncoder subclass: #EncoderForSistaV1 (excessive size, no diff calculated) Item was added: + ----- Method: EncoderForSistaV1>>genSendDirectedSuper:numArgs: (in category 'bytecode generation') ----- + genSendDirectedSuper: selectorLiteralIndex numArgs: nArgs + | extendedIndex | + (selectorLiteralIndex < 0 or: [selectorLiteralIndex > 65535]) ifTrue: + [^self outOfRangeError: 'selectorLiteralIndex' index: selectorLiteralIndex range: 0 to: 65535]. + (nArgs < 0 or: [nArgs > 31]) ifTrue: + [^self outOfRangeError: 'numArgs' index: nArgs range: 0 to: 31 "!!!!"]. + (extendedIndex := selectorLiteralIndex) > 31 ifTrue: + [self genUnsignedSingleExtendA: extendedIndex // 32. + extendedIndex := extendedIndex \\ 32]. + "Bit 6 of the ExtB byte is the directed send flag. Bit 6 allows for future expansion to up to 255 args." + self genUnsignedSingleExtendB: nArgs // 8 + 64. + "235 11101011 iiiiijjj Send To Superclass Literal Selector #iiiii (+ Extend A * 32) with jjj (+ Extend B * 8) Arguments" + stream + nextPut: 235; + nextPut: nArgs \\ 8 + (extendedIndex * 8)!

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 April 2015