Vm-dev April 2015

vm-dev@lists.squeakfoundation.org

45 participants
264 discussions

About primitiveBitAnd / Or / Xor
by Nicolas Cellier 01 Apr '15

01 Apr '15

Hi again, I've noticed the COG code currently branches on objectMemory wordSize = 8 condition, to either load self positive64BitValueOf: integerArgument. or to load self positive32BitValueOf: integerArgument. Of course, this could be resolved to a single integerArgument := self positiveMachineIntegerValueOf: integerArgument which will care of the branch by itself... That's what I did. But maybe the intention was rather to deal with signed64BitValueOf in the 64 bit branch like what was done to primitiveBitShift? Of course, all this is assuming that the machine uses a 2-complement implementation, otherwise it won't match the Smalltalk semantic. That's not so rare ;) - but theoretically not fully portable - contrarily to actual strategy which is fully portable. Since other parts of the VM also rely on such specific, maybe we can trade a little portability tribute for the sake of speed. Thoughts? Nicolas

2 2

Re: [Vm-dev] VM Maker: VMMaker.oscog-eem.1146.mcz
by Tobias Pape 01 Apr '15

01 Apr '15

On 01.04.2015, at 18:37, commits(a)source.squeak.org wrote: > Eliot Miranda uploaded a new version of VMMaker to project VM Maker: > http://source.squeak.org/VMMaker/VMMaker.oscog-eem.1146.mcz > > ==================== Summary ==================== > > Name: VMMaker.oscog-eem.1146 > Author: eem > Time: 1 April 2015, 11:35:11.604 am > UUID: 0bdb63f1-8a30-4a34-9300-43937b9b3aa5 > Ancestors: VMMaker.oscog-eem.1145 > > ...and fix primitiveIsBigEnder for Cog. > Care to explain briefly? > =============== Diff against VMMaker.oscog-eem.1145 =============== > > Item was changed:

2 2

VM Maker: VMMaker.oscog-eem.1146.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.1146.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.1146 Author: eem Time: 1 April 2015, 11:35:11.604 am UUID: 0bdb63f1-8a30-4a34-9300-43937b9b3aa5 Ancestors: VMMaker.oscog-eem.1145 ...and fix primitiveIsBigEnder for Cog. =============== Diff against VMMaker.oscog-eem.1145 =============== Item was changed: ----- Method: InterpreterPrimitives>>primitiveIsBigEnder (in category 'other primitives') ----- primitiveIsBigEnder "Answer if running on a big endian machine." <export: true> + self pop: 1 thenPushBool: VMBIGENDIAN! - self pop: 1 thenPushBool: self isBigEnder!

1 0

VM Maker: VMMaker.oscog-eem.1145.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.1145.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.1145 Author: eem Time: 1 April 2015, 11:04:56.527 am UUID: 945d0a43-e29d-4fb4-a1b8-263465636aec Ancestors: VMMaker.oscog-eem.1144 Retract the decision to type the results of the bitwise operators as unsigned; it breaks at least the BitBlt plugin. =============== Diff against VMMaker.oscog-tpr.1143 =============== Item was changed: ----- Method: CCodeGenerator>>returnTypeForSend:in: (in category 'type inference') ----- returnTypeForSend: sendNode in: aTMethod "Answer the return type for a send. Absent sends default to #sqInt. The bitwise operators answer unsigned versions of their argument types, at least in gcc although this author can't find that in the C99 spec. If you can find this, please let me know." | sel | ^(self anyMethodNamed: (sel := sendNode selector)) ifNil: [kernelReturnTypes at: sel ifAbsent: [^sel caseOf: { [#+] -> [self typeForArithmetic: sendNode in: aTMethod]. [#-] -> [self typeForArithmetic: sendNode in: aTMethod]. [#*] -> [self typeForArithmetic: sendNode in: aTMethod]. [#/] -> [self typeForArithmetic: sendNode in: aTMethod]. [#addressOf:] -> [(self typeFor: sendNode receiver in: aTMethod) ifNil: [#sqInt] ifNotNil: [:type| type, (type last isLetter ifTrue: [' *'] ifFalse: ['*'])]]. [#at:] -> [self typeForDereference: sendNode in: aTMethod]. + [#bitAnd:] -> [self typeForArithmetic: sendNode in: aTMethod]. + [#bitOr:] -> [self typeForArithmetic: sendNode in: aTMethod]. + [#bitXor:] -> [self typeForArithmetic: sendNode in: aTMethod]. - [#bitAnd:] -> [self unsignedTypeForBitwiseSend: sendNode in: aTMethod]. - [#bitOr:] -> [self unsignedTypeForBitwiseSend: sendNode in: aTMethod]. - [#bitXor:] -> [self unsignedTypeForBitwiseSend: sendNode in: aTMethod]. [#asFloat] -> [#double]. [#atan] -> [#double]. [#exp] -> [#double]. [#log] -> [#double]. [#sin] -> [#double]. [#sqrt] -> [#double]. [#asLong] -> [#long]. [#asUnsignedInteger] -> [#usqInt]. [#asUnsignedLong] -> [#'unsigned long']. [#asVoidPointer] -> [#'void *']. [#signedIntToLong] -> [#usqInt]. "c.f. generateSignedIntToLong:on:indent:" [#signedIntToShort] -> [#usqInt]. "c.f. generateSignedIntToShort:on:indent:" [#cCoerce:to:] -> [sendNode args last value]. [#cCoerceSimple:to:] -> [sendNode args last value]. [#ifTrue:ifFalse:] -> [self typeForConditional: sendNode in: aTMethod]. [#ifFalse:ifTrue:] -> [self typeForConditional: sendNode in: aTMethod]. [#ifTrue:] -> [self typeForConditional: sendNode in: aTMethod]. [#ifFalse:] -> [self typeForConditional: sendNode in: aTMethod] } otherwise: [#sqInt]]] ifNotNil: [:m| m returnType ifNotNil: [:type| self baseTypeForType: type]]! Item was removed: - ----- Method: CCodeGenerator>>unsignedTypeForBitwiseSend:in: (in category 'type inference') ----- - unsignedTypeForBitwiseSend: aTSendNode in: aTMethod - "The result of the bitwise operators in C is unsigned. - We make an exception for currentBytecode bitAnd: N to allow better inlining in cases." - | t1 t2 type | - (aTSendNode receiver isVariable - and: [aTSendNode receiver name = 'currentBytecode' - and: [aTSendNode selector == #bitAnd:]]) ifTrue: - [^#sqInt]. - t1 := (self typeFor: aTSendNode receiver in: aTMethod) ifNil: [^nil]. - t2 := (self typeFor: aTSendNode args first in: aTMethod) ifNil: [^nil]. - type := (self sizeOfIntegralCType: t1) >= (self sizeOfIntegralCType: t1) - ifTrue: [t1] - ifFalse: [t2]. - ^type first = $u - ifTrue: [type] - ifFalse: - [(type beginsWith: 'sq') - ifTrue: ['u', type] - ifFalse: ['unsigned ', type]]! Item was changed: ----- Method: InterpreterPrimitives>>primitiveAdoptInstance (in category 'object access primitives') ----- primitiveAdoptInstance "Primitive. Change the class of the argument to make it an instance of the receiver given that the format of the receiver matches the format of the argument's class. Fail if receiver or argument are SmallIntegers, or the receiver is an instance of a compact class and the argument isn't, or when the argument's class is compact and the receiver isn't, or when the format of the receiver is different from the format of the argument's class, or when the arguments class is fixed and the receiver's size differs from the size that an instance of the argument's class should have." | rcvr arg err | arg := self stackObjectValue: 0. rcvr := self stackObjectValue: 1. + self successful ifFalse:[^nil]. err := objectMemory changeClassOf: arg to: rcvr. err = 0 ifTrue: ["Flush at cache because rcvr's class has changed." self flushAtCache. self pop: self methodArgumentCount] ifFalse: [self primitiveFailFor: err]. ^nil! Item was added: + ----- Method: InterpreterPrimitives>>primitiveIsBigEnder (in category 'other primitives') ----- + primitiveIsBigEnder + "Answer if running on a big endian machine." + <export: true> + self pop: 1 thenPushBool: self isBigEnder! Item was changed: ----- Method: NSSendCacheSurrogate32>>enclosingObject: (in category 'accessing') ----- enclosingObject: aValue ^memory unsignedLongAt: address + 13 put: aValue!

1 0

VM Maker: BytecodeSets.spur-cb.24.mcz
by commits＠source.squeak.org 01 Apr '15

01 Apr '15

ClementBera uploaded a new version of BytecodeSets to project VM Maker: http://source.squeak.org/VMMaker/BytecodeSets.spur-cb.24.mcz ==================== Summary ==================== Name: BytecodeSets.spur-cb.24 Author: cb Time: 1 April 2015, 10:35:05.836 am UUID: 552f5804-99ea-47b6-ab5c-a6556e7a9cc3 Ancestors: BytecodeSets.spur-cb.23 fix two encodings bugs: - popIntoArray not using the correct byte - some sends not using the correct byte. =============== Diff against BytecodeSets.spur-cb.23 =============== Item was changed: ----- Method: EncoderForSistaV1>>genPushConsArray: (in category 'bytecode generation') ----- genPushConsArray: size (size < 0 or: [size > 127]) ifTrue: [^self outOfRangeError: 'size' index: size range: 0 to: 127]. + "231 11100111 jkkkkkkk Push (Array new: kkkkkkk) (j = 0) - "233 11101001 jkkkkkkk Push (Array new: kkkkkkk) (j = 0) & Pop kkkkkkk elements into: (Array new: kkkkkkk) (j = 1)" stream + nextPut: 231; - nextPut: 233; nextPut: size + 128! Item was changed: ----- Method: EncoderForSistaV1>>genSend:numArgs: (in category 'bytecode generation') ----- genSend: selectorLiteralIndex numArgs: nArgs | extendedIndex extendedNArgs | (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 "!!!!"]. (selectorLiteralIndex < 16 and: [nArgs < 3]) ifTrue: ["128-143 1000 iiii Send Literal Selector #iiii With 0 Argument 144-159 1001 iiii Send Literal Selector #iiii With 1 Arguments 160-175 1010 iiii Send Literal Selector #iiii With 2 Arguments" stream nextPut: 128 + (nArgs * 16) + selectorLiteralIndex. ^self]. (extendedIndex := selectorLiteralIndex) > 31 ifTrue: [self genUnsignedSingleExtendA: extendedIndex // 32. extendedIndex := extendedIndex \\ 32]. (extendedNArgs := nArgs) > 7 ifTrue: [self genUnsignedSingleExtendB: extendedNArgs // 8. extendedNArgs := extendedNArgs \\ 8]. "234 11101010 i i i i i j j j Send Literal Selector #iiiii (+ Extend A * 32) with jjj (+ Extend B * 8) Arguments" stream + nextPut: 234; - nextPut: 238; nextPut: extendedNArgs + (extendedIndex * 8)!

1 0

VM Maker: BytecodeSets.spur-cb.23.mcz
by commits＠source.squeak.org 01 Apr '15

01 Apr '15

ClementBera uploaded a new version of BytecodeSets to project VM Maker: http://source.squeak.org/VMMaker/BytecodeSets.spur-cb.23.mcz ==================== Summary ==================== Name: BytecodeSets.spur-cb.23 Author: cb Time: 1 April 2015, 10:02:59.786 am UUID: dd28463a-4917-4cef-9188-a457409fe2c2 Ancestors: BytecodeSets.spur-cb.22 Fixed bug storeTemp was compiled as storePopTemp. Fix a method comment. Fix selectorToSendOrSelf which call an inexisting method. The image running with the new bytecode set still crashes (but it can also be a VM side bug). =============== Diff against BytecodeSets.spur-cb.22 =============== Item was changed: ----- Method: EncoderForSistaV1 class>>selectorToSendOrItselfFor:in:at: (in category 'instruction stream support') ----- selectorToSendOrItselfFor: anInstructionStream in: method at: pc "If anInstructionStream is at a send bytecode then answer the send's selector, otherwise answer anInstructionStream itself. The rationale for answering anInstructionStream instead of, say, nil, is that potentially any existing object can be used as a selector, but since anInstructionStream postdates the method, it can't be one of them. The compilcation is that for convenience we assume the pc could be pointing to the raw send bytecode after its extensions, or at the extension preceeding the raw send bytecode. 96-111 0110 iiii Send Arithmetic Message #iiii #(#+ #- #< #> #'<=' #'>=' #= #'~=' #* #/ #'\\' #@ #bitShift: #'//' #bitAnd: #bitOr:) 112-119 01110 iii Send Special Message #iii #(#at: #at:put: #size #next #nextPut: #atEnd #'==' class) 120 01111000 UNASSIGNED (was: blockCopy:) 121 01111001 Send Special Message #value 122-123 0111101 i Send Special Message #i #(#value: #do:) 124-127 011111 ii Send Special Message #ii #(#new #new: #x #y)) 128-143 1000 iiii Send Literal Selector #iiii With 0 Argument 144-159 1001 iiii Send Literal Selector #iiii With 1 Arguments 160-175 1010 iiii Send Literal Selector #iiii With 2 Arguments * 224 11100000 aaaaaaaa Extend A (Ext A = Ext A prev * 256 + Ext A) * 225 11100001 bbbbbbbb Extend B (Ext B = Ext B prev * 256 + Ext B) ** 234 11101010 iiiiijjj Send Literal Selector #iiiii (+ Extend A * 32) with jjj (+ Extend B * 8) Arguments ** 235 11101011 iiiiijjj Send To Superclass Literal Selector #iiiii (+ Extend A * 32) with jjj (+ Extend B * 8) Arguments" | byte | byte := method at: pc. byte < 96 ifTrue: [^anInstructionStream]. byte <= 175 ifTrue: ["special byte or short send" ^byte >= 128 ifTrue: [method literalAt: (byte bitAnd: 15) + 1] ifFalse: [Smalltalk specialSelectorAt: byte - 95]]. byte < 234 ifTrue: "need to check for either extension cuz order of extensions is not restricted. so extB could preceed extA" [(byte >= 224 and: [byte <= 225]) ifTrue: [^self extensionsAt: pc in: method into: [:extA :extB :nExtBytes| | byteAfter index | byteAfter := method at: pc + nExtBytes. (byteAfter >= 234 and: [byteAfter <= 235]) ifTrue: [index := ((method at: pc + nExtBytes + 1) bitShift: -3) + (extA bitShift: 5). method literalAt: index + 1] ifFalse: [anInstructionStream]]]. ^anInstructionStream]. byte > 235 ifTrue: [^anInstructionStream]. "they could be extended..." + ^self extensionsAt: pc in: method into: - ^self extensionsFor: pc in: method into: [:extA :extB :nExtBytes| | index | index := ((method at: pc + 1) bitShift: -3) + (extA bitShift: 5). method literalAt: index + 1]! Item was changed: ----- Method: EncoderForSistaV1>>genReturnTopToCaller (in category 'bytecode generation') ----- genReturnTopToCaller + "94 01011110 Return Stack Top From Block [* return from enclosing block N, ExtA]" - "94 1011101 Return Stack Top From Block [* return from enclosing block N, ExtA]" "If extended, the least significant bit of the extension determines if we return to the caller or not and the most significant bits determine how many levels of the static chain to return from. ExtA = iiiiiiij iiiiiii=0,j=0 => return to caller iiiiiii=0,j=1 => illegal iiiiiii=1,j=0 => return to outerContext iiiiiii=1,j=1 => return to outerContext sender/return from outerContext iiiiiii=2,j=0 => return to outerContext outerContext iiiiiii=2,j=1 => return to outerContext outerContext sender/return from outerContext outerContext etc" stream nextPut: 94! Item was changed: ----- Method: EncoderForSistaV1>>genStoreTemp: (in category 'bytecode generation') ----- genStoreTemp: tempIndex + "245 11110110 iiiiiiii Store Temporary Variable #iiiiiiii" - "242 11110010 iiiiiiii Pop and Store Temporary Variable #iiiiiiii" (tempIndex < 0 or: [tempIndex > 63]) ifTrue: [^self outOfRangeError: 'index' index: tempIndex range: 0 to: 63]. stream + nextPut: 245; - nextPut: 242; nextPut: tempIndex!

1 0

VM Maker: BytecodeSets.spur-cb.22.mcz
by commits＠source.squeak.org 01 Apr '15

01 Apr '15

ClementBera uploaded a new version of BytecodeSets to project VM Maker: http://source.squeak.org/VMMaker/BytecodeSets.spur-cb.22.mcz ==================== Summary ==================== Name: BytecodeSets.spur-cb.22 Author: cb Time: 1 April 2015, 9:40:25.46 am UUID: 23559a1a-1471-49b6-b888-477e71431fe7 Ancestors: BytecodeSets.spur-cb.21 empty log message =============== Diff against BytecodeSets.spur-cb.21 =============== Item was changed: ----- Method: EncoderForSistaV1 class>>selectorToSendOrItselfFor:in:at: (in category 'instruction stream support') ----- selectorToSendOrItselfFor: anInstructionStream in: method at: pc "If anInstructionStream is at a send bytecode then answer the send's selector, otherwise answer anInstructionStream itself. The rationale for answering anInstructionStream instead of, say, nil, is that potentially any existing object can be used as a selector, but since anInstructionStream postdates the method, it can't be one of them. The compilcation is that for convenience we assume the pc could be pointing to the raw send bytecode after its extensions, or at the extension preceeding the raw send bytecode. 96-111 0110 iiii Send Arithmetic Message #iiii #(#+ #- #< #> #'<=' #'>=' #= #'~=' #* #/ #'\\' #@ #bitShift: #'//' #bitAnd: #bitOr:) 112-119 01110 iii Send Special Message #iii #(#at: #at:put: #size #next #nextPut: #atEnd #'==' class) 120 01111000 UNASSIGNED (was: blockCopy:) 121 01111001 Send Special Message #value 122-123 0111101 i Send Special Message #i #(#value: #do:) 124-127 011111 ii Send Special Message #ii #(#new #new: #x #y)) 128-143 1000 iiii Send Literal Selector #iiii With 0 Argument 144-159 1001 iiii Send Literal Selector #iiii With 1 Arguments 160-175 1010 iiii Send Literal Selector #iiii With 2 Arguments * 224 11100000 aaaaaaaa Extend A (Ext A = Ext A prev * 256 + Ext A) * 225 11100001 bbbbbbbb Extend B (Ext B = Ext B prev * 256 + Ext B) ** 234 11101010 iiiiijjj Send Literal Selector #iiiii (+ Extend A * 32) with jjj (+ Extend B * 8) Arguments ** 235 11101011 iiiiijjj Send To Superclass Literal Selector #iiiii (+ Extend A * 32) with jjj (+ Extend B * 8) Arguments" | byte | byte := method at: pc. byte < 96 ifTrue: [^anInstructionStream]. byte <= 175 ifTrue: ["special byte or short send" ^byte >= 128 ifTrue: [method literalAt: (byte bitAnd: 15) + 1] + ifFalse: [Smalltalk specialSelectorAt: byte - 95]]. - ifFalse: [Smalltalk specialSelectorAt: byte - 79]]. byte < 234 ifTrue: "need to check for either extension cuz order of extensions is not restricted. so extB could preceed extA" [(byte >= 224 and: [byte <= 225]) ifTrue: [^self extensionsAt: pc in: method into: [:extA :extB :nExtBytes| | byteAfter index | byteAfter := method at: pc + nExtBytes. (byteAfter >= 234 and: [byteAfter <= 235]) ifTrue: [index := ((method at: pc + nExtBytes + 1) bitShift: -3) + (extA bitShift: 5). method literalAt: index + 1] ifFalse: [anInstructionStream]]]. ^anInstructionStream]. byte > 235 ifTrue: [^anInstructionStream]. "they could be extended..." ^self extensionsFor: pc in: method into: [:extA :extB :nExtBytes| | index | index := ((method at: pc + 1) bitShift: -3) + (extA bitShift: 5). method literalAt: index + 1]! Item was changed: ----- Method: EncoderForSistaV1>>genReturnTopToCaller (in category 'bytecode generation') ----- genReturnTopToCaller + "94 1011101 Return Stack Top From Block [* return from enclosing block N, ExtA]" - "93 1011101 Return Stack Top From Block [* return from enclosing block N, ExtA]" "If extended, the least significant bit of the extension determines if we return to the caller or not and the most significant bits determine how many levels of the static chain to return from. ExtA = iiiiiiij iiiiiii=0,j=0 => return to caller iiiiiii=0,j=1 => illegal iiiiiii=1,j=0 => return to outerContext iiiiiii=1,j=1 => return to outerContext sender/return from outerContext iiiiiii=2,j=0 => return to outerContext outerContext iiiiiii=2,j=1 => return to outerContext outerContext sender/return from outerContext outerContext etc" stream nextPut: 94!

1 0

Re: [squeak-dev] Bug whith contextOn:do:
by Eliot Miranda 01 Apr '15

01 Apr '15

Hi Tobias, first let me sympathize. This is such horrible code :-(. Ovr the years, getting this code to work with Cog's context-to-stack-mapping machinery has given me headaches :-). On Tue, Mar 31, 2015 at 9:01 AM, Tobias Pape <Das.Linux(a)gmx.de> wrote: > Hey fellow Squeakers > > [Attention: low level lengthy stuff] > > I today encountered a strange behavior. > When running the Startup code, somewhen ContextPart>>#complete is called, > which, in the process issues #contextOn:do:, which subsequently somewhere > interanlly > does a jump: > > contextOn: exceptionClass do: block > "Create an #on:do: context that is ready to return from executing > its receiver" > > | ctxt chain | > ctxt := thisContext. > [chain := thisContext sender cut: ctxt. ctxt jump] on: > exceptionClass do: block. > "jump above will resume here without unwinding chain" > ^ chain > > The idea is that we end up right before '^ chain' as the comment indicates. > so what happens is that ctxt is the context of #contextOn:do: itself and > it gets > send #jump in the closure. > > Jump now does the following: > > jump > "Abandon thisContext and resume self instead (using the same > current process). You may want to save thisContext's sender before calling > this so you can jump back to it. > Self MUST BE a top context (ie. a suspended context or a abandoned > context that was jumped out of). A top context already has its return > value on its stack (see Interpreter>>primitiveSuspend and other suspending > primitives). > thisContext's sender is converted to a top context (by pushing a > nil return value on its stack) so it can be jump back to." > > | top | > "Make abandoned context a top context (has return value (nil)) so > it can be jumped back to" > thisContext sender push: nil. > > "Pop self return value then return it to self (since we jump to > self by returning to it)" > stackp = 0 ifTrue: [self stepToSendOrReturn]. > stackp = 0 ifTrue: [self push: nil]. "must be quick return > self/constant" > top := self pop. > thisContext privSender: self. > ^ top > > So. bytecode for #contextOn:do: is: > > 29 <8A 01> push: (Array new: 1) > 31 <6B> popIntoTemp: 3 > 32 <89> pushThisContext: > 33 <6A> popIntoTemp: 2 > 34 <12> pushTemp: 2 > 35 <13> pushTemp: 3 > 36 <8F 20 00 0A> closureNumCopied: 2 numArgs: 0 bytes 40 to 49 > 40 <89> pushThisContext: > 41 <D2> send: sender > 42 <10> pushTemp: 0 > 43 <E1> send: cut: > 44 <8E 00 01> popIntoTemp: 0 inVectorAt: 1 > 47 <10> pushTemp: 0 > 48 <D3> send: jump > 49 <7D> blockReturn > 50 <10> pushTemp: 0 > 51 <11> pushTemp: 1 > 52 <F0> send: on:do: > 53 <87> pop > 54 <8C 00 03> pushTemp: 0 inVectorAt: 3 > 57 <7C> returnTop > > > The jump lands right at 53 and does a pop. > HOWEVER, at this point the stack of this context is empty and the pop > actually pops the 3rd temp > from the temps that 'just happens' to be right under the stack. This > should be fatal. > HOWEVER again, Squeak actually does not pop but only decrement the SP so > the temp access still > works(this _could_ be fine but some implementations (Eg, RSqueak) tried > to separate temps and > stack; which is not possible currently). > > What could be the problem here? > - are the 'stackp = 0'-checks in #jump wrong and they actually should > check for the actual stack depth _after_ temps? > It does look like it. I would have expected this to be more correct: jump "Abandon thisContext and resume self instead (using the same current process). You may want to save thisContext's sender before calling this so you can jump back to it. Self MUST BE a top context (ie. a suspended context or a abandoned context that was jumped out of). A top context already has its return value on its stack (see Interpreter>>primitiveSuspend and other suspending primitives). thisContext's sender is converted to a top context (by pushing a nil return value on its stack) so it can be jump back to." | top | "Make abandoned context a top context (has return value (nil)) so it can be jumped back to" thisContext sender push: nil. "Pop self return value then return it to self (since we jump to self by returning to it)" stackp <= self numTemps ifTrue: [self stepToSendOrReturn]. (stackp <= self numTemps and: [self willJustPop]) ifTrue: [self push: nil]. "must be quick return self/constant" top := self pop. thisContext privSender: self. ^top > - should we put in a "sacrificial anode" in #contextOn:do: so that the pop > does not pop the empty stack? (like this: > > contextOn: exceptionClass do: block > "Create an #on:do: context that is ready to return from executing > its receiver" > > | ctxt chain | > ctxt := thisContext. > [chain := thisContext sender cut: ctxt. > ctxt push: nil. "sacrifical anode" > ctxt jump > ] on: exceptionClass do: block. > "jump above will resume here without unwinding chain" > ^ chain > That looks right. Once the on:do: is sent the thisContext of contextOn:do:'s stack contains only exceptionClass, block, context and the indirection vector containing chain. So it is in the stack = self numTemps case. > > - Or is there an even better way? > I'm not sure the other ways are any better. The way to transfer to a context without disturbing its stack is to do a process switch. So you do something equivalent to jump "Abandon thisContext and resume self instead (using the same current process)." | process semaphore | process := Processor activeProcess. semaphore := Semaphore new. [process suspendedContext unwindTo: self. process suspendedContext: self. semaphore signal] fork. semaphore wait This way no bizarre stack manipulations are going on, and no return value is pushed, because there isn't a return. One may get away with: jump "Abandon thisContext and resume self instead (using the same current process)." [| process | process := Processor activeProcess. process suspendedContext unwindTo: self. process suspendedContext: self] fork. Processor yield I'd be interested in your experience using either of these. One of the advantages the process switch versions have is in not updating the receiving context sp there's a chance the context-to-stack mapping machinery won't flush the context to the heap. In the end it might actually be faster. -- best, Eliot

2 2

Re: [Vm-dev] VM Maker: BytecodeSets.spur-cb.21.mcz
by Levente Uzonyi 01 Apr '15

01 Apr '15

I guess the comment should be updated to match the change of the code. Levente On Wed, 1 Apr 2015, commits(a)source.squeak.org wrote: > > ClementBera uploaded a new version of BytecodeSets to project VM Maker: > http://source.squeak.org/VMMaker/BytecodeSets.spur-cb.21.mcz > > ==================== Summary ==================== > > Name: BytecodeSets.spur-cb.21 > Author: cb > Time: 1 April 2015, 9:07:51.823 am > UUID: 2b6da806-8ef3-4f4c-acb6-28d7baffe999 > Ancestors: BytecodeSets.spur-eem.20 > > empty log message > > =============== Diff against BytecodeSets.spur-eem.20 =============== > > Item was changed: > ----- Method: EncoderForSistaV1>>genReturnTopToCaller (in category 'bytecode generation') ----- > genReturnTopToCaller > "93 1011101 Return Stack Top From Block [* return from enclosing block N, ExtA]" > "If extended, the least significant bit of the extension determines if we return to the caller or not > and the most significant bits determine how many levels of the static chain to return from. > ExtA = iiiiiiij > iiiiiii=0,j=0 => return to caller > iiiiiii=0,j=1 => illegal > iiiiiii=1,j=0 => return to outerContext > iiiiiii=1,j=1 => return to outerContext sender/return from outerContext > iiiiiii=2,j=0 => return to outerContext outerContext > iiiiiii=2,j=1 => return to outerContext outerContext sender/return from outerContext outerContext > etc" > > + stream nextPut: 94! > - stream nextPut: 93! > >

1 0

VM Maker: BytecodeSets.spur-cb.21.mcz
by commits＠source.squeak.org 01 Apr '15

01 Apr '15

ClementBera uploaded a new version of BytecodeSets to project VM Maker: http://source.squeak.org/VMMaker/BytecodeSets.spur-cb.21.mcz ==================== Summary ==================== Name: BytecodeSets.spur-cb.21 Author: cb Time: 1 April 2015, 9:07:51.823 am UUID: 2b6da806-8ef3-4f4c-acb6-28d7baffe999 Ancestors: BytecodeSets.spur-eem.20 empty log message =============== Diff against BytecodeSets.spur-eem.20 =============== Item was changed: ----- Method: EncoderForSistaV1>>genReturnTopToCaller (in category 'bytecode generation') ----- genReturnTopToCaller "93 1011101 Return Stack Top From Block [* return from enclosing block N, ExtA]" "If extended, the least significant bit of the extension determines if we return to the caller or not and the most significant bits determine how many levels of the static chain to return from. ExtA = iiiiiiij iiiiiii=0,j=0 => return to caller iiiiiii=0,j=1 => illegal iiiiiii=1,j=0 => return to outerContext iiiiiii=1,j=1 => return to outerContext sender/return from outerContext iiiiiii=2,j=0 => return to outerContext outerContext iiiiiii=2,j=1 => return to outerContext outerContext sender/return from outerContext outerContext etc" + stream nextPut: 94! - stream nextPut: 93!

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