Vm-dev May 2021

vm-dev@lists.squeakfoundation.org

21 participants
56 discussions

walkback while generating external plugins
by Robert Withers 27 May '21

27 May '21

Hi, I am getting a walkback when generating new external plugins: RSErasurePlugin and RSFECPlugin. The culprit seems to be a super call to InterpreterPlugin class>>#declareCVarsIn:, called from RSErasurePlugin class>>declareCVarsIn:. I snipped the stack to pertinent frames. Thx! 27 May 2021 7:44:38.719538 am VM: unix - Smalltalk Image: Squeak6.0alpha [latest update: #20535] SecurityManager state: Restricted: false FileAccess: true SocketAccess: true Working Dir /home/rabbit/vm-dev/opensmalltalk-vm/image Trusted Dir /home/rabbit/vm-dev/opensmalltalk-vm/image/secure Untrusted Dir /home/rabbit/vm-dev/opensmalltalk-vm/image/My Squeak ByteSymbol(Object)>>doesNotUnderstand: #removeVariable:ifAbsent: Receiver: #cg Arguments and temporary variables: aMessage: removeVariable: 'translatedMethodCache' ifAbsent: nil exception: MessageNotUnderstood: ByteSymbol>>removeVariable:ifAbsent: resumeValue: nil Receiver's instance variables: #cg RSErasurePlugin class(InterpreterPlugin class)>>declareCVarsIn: Receiver: RSErasurePlugin Arguments and temporary variables: aCCodeGenerator: #cg Receiver's instance variables: superclass: InterpreterPlugin methodDict: a MethodDictionary(#add:by:->(RSErasurePlugin>>#add:by: "a CompiledMethod...etc... format: 65542 instanceVariables: #('logTable' 'expTable' 'multiplicationTable') organization: ('arithmetic' add:by: divide:by: exp:power: galoisMultiply:by: subtractFrom:by:...etc... subclasses: {RSErasurePluginSimulator} name: #RSErasurePlugin classPool: nil sharedPools: nil environment: Smalltalk category: #CryptographyRSFECPlugin timeStamp: 3799545237 RSErasurePlugin class>>declareCVarsIn: Receiver: RSErasurePlugin Arguments and temporary variables: cg: a VMPluginCodeGenerator Receiver's instance variables: superclass: InterpreterPlugin methodDict: a MethodDictionary(#add:by:->(RSErasurePlugin>>#add:by: "a CompiledMethod...etc... format: 65542 instanceVariables: #('logTable' 'expTable' 'multiplicationTable') organization: ('arithmetic' add:by: divide:by: exp:power: galoisMultiply:by: subtractFrom:by:...etc... subclasses: {RSErasurePluginSimulator} name: #RSErasurePlugin classPool: nil sharedPools: nil environment: Smalltalk category: #CryptographyRSFECPlugin timeStamp: 3799545237 -- --- Kindly, Robert

2 3

#doesNotUnderstand: in BuildSqueakSpurTrunkVMMakerImage.st
by Robert Withers 27 May '21

27 May '21

Hello everyone. I hope you all are well. I am trying to build a VMMaker image to generate plugins for the new ReedSolomon Erasure coding and Error Correction, in Cryptography. I cloned the opensmalltalk-vm and am running ./buildspurtrunkvmmaker64image.sh -vm /usr/local/bin/squeakthe ''. When it gets to building the BuildSqueakSpurTrunkVMMakerImage I receive a DNU regarding external function. I have attached the stack trace. /usr/local/bin/squeak Spur64VMMaker.image BuildSqueakSpurTrunkVMMakerImage.st I appreciate your assistance! -- Kindly, Robert

5 6

new cog-spur VM for OpenIndiana
by stes＠PANDORA.BE 23 May '21

23 May '21

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 Hello, Update about the OpenSmalltalk VM on "OpenIndiana". See docs and screenshots of Squeak and Cuis on "OpenIndiana" at: http://docs.openindiana.org/handbook/community/squeak/index.html I have renamed the packages to squeak, stack-spur and cog-spur respectively. The version number 5.0.2957 is something of my own fabrication, based on the names VMMaker.oscog-eem.2957 for Stack and CoInterpreter. # pkg info cog-spur Name: runtime/smalltalk/cog-spur Summary: The OpenSmalltalk Cog Spur Virtual Machine Category: Development/Smalltalk State: Installed Publisher: userland Version: 5.0.2957 Branch: 2020.0.1.0 Packaging Date: Sat May 15 10:35:19 2021 Last Install Time: Sat May 15 10:46:22 2021 Size: 1.00 MB Thanks, David Stes -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQEcBAEBCAAGBQJgqqPJAAoJENdFDkXGicizquQIAMX3NQcUaDXyT3wYOQ6M+lJb i3h8w9PDDKVRdj4Gys3YnZwQSl6omtgd9IjthcM9bQ3CrayVb23WJOGMal+7dpEk 5OIXELGb/a6/uzc48FNJIbyw4b8kB3XcZYyflCJYEHGQX0/eteBl7kBPWjiXL0EI G2ZOkI2NnZl/1JM2/3kLVQG0o+P1MqOW0pZ5m+rsSbhOP5cJYUaLdSW7CgiWDbpd +mTWo9RspAHRrcVIjMiKCvKded8zCxLg+Pwdk3efk+PPzcnzUjYOxi2ZJ9mPDFLB hdp+S5XHJ+CMMgmMHrPTmc5L8b7VFAuo/KbSJerIXciqios4u9rRscNDYiKguIg= =I/Sm -----END PGP SIGNATURE-----

1 0

VM Maker: VMMaker.oscog-eem.2962.mcz
by commits＠source.squeak.org 22 May '21

22 May '21

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.2962.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2962 Author: eem Time: 21 May 2021, 3:22:33.156394 pm UUID: 8b5a2a1e-46d4-445a-a0eb-95c71ffc7e16 Ancestors: VMMaker.oscog-eem.2961 Implement as intended (i.e. treat the parameter as an iomage-level Smalltalk object) InterpreterProxy>>isOopMutable:/isOopImmutable: =============== Diff against VMMaker.oscog-eem.2961 =============== Item was changed: + ----- Method: InterpreterProxy>>isOopImmutable: (in category 'testing') ----- - ----- Method: InterpreterProxy>>isOopImmutable: (in category 'object access') ----- isOopImmutable: oop <option: #IMMUTABILITY> + ^oop isReadOnlyObject! - ^StackInterpreter objectMemoryClass isOopImmutable: oop! Item was changed: ----- Method: InterpreterProxy>>isOopMutable: (in category 'testing') ----- isOopMutable: anOop <api> + ^anOop isReadOnlyObject not! - ^self notYetImplementedError!

1 0

Intermittent AttachableFileStream error during 7:36 and 10:36 today
by Bob Westergaard 21 May '21

21 May '21

Hi, I have a rather bizarre intermittent failure that I'd see if anyone else can reproduce. I've taken the following: http://files.squeak.org/5.3/Squeak5.3-19435-64bit/Squeak5.3-19435-64bit-202… Started it up and loaded the latest system updates and the OSProcess package (access OS functions). I then did: CommandShell command: 'date' This works as I'd expect it too. However, what is interesting is that if I change the time on the machine to any time between 7:36AM and 10:36AM (PDT, I'm in San Jose, CA) for today (May 19th) this call fails with the stack below. I'm quitting the image between time changes. There was no failure for the same time period yesterday. Nor does it appear to fail for any other time period today. -- Bob Stack follows: 19 May 2021 10:30:20.503624 am VM: unix - Smalltalk Image: Squeak5.3 [latest update: #19458] SecurityManager state: Restricted: false FileAccess: true SocketAccess: true Working Dir /scratch/bwesterg/squeak/Squeak5.3-19435-64bit-202003021730-Linux/shared Trusted Dir /scratch/bwesterg/squeak/Squeak5.3-19435-64bit-202003021730-Linux/shared/secure Untrusted Dir /scratch/bwesterg/squeak/Squeak5.3-19435-64bit-202003021730-Linux/shared/My Squeak AioEventHandler(Object)>>notify: Receiver: an AioEventHandler Arguments and temporary variables: aString: 'aio event forwarding not supported' Receiver's instance variables: dependents: nil semaphore: nil semaIndex: nil handlerProc: nil descriptor: nil [] in AioEventHandler>>initializeForExceptions:readEvents:writeEvents: Receiver: an AioEventHandler Arguments and temporary variables: <<error during printing> Receiver's instance variables: dependents: nil semaphore: nil semaIndex: nil handlerProc: nil descriptor: nil BlockClosure>>cull: Receiver: [closure] in AioEventHandler>>initializeForExceptions:readEvents:writeEvents: Arguments and temporary variables: firstArg: Warning: aio event forwarding not supported Receiver's instance variables: outerContext: AioEventHandler>>initializeForExceptions:readEvents:writeEvents: startpc: 154 numArgs: 1 [] in Context>>handleSignal: Receiver: BlockClosure>>on:do: Arguments and temporary variables: <<error during printing> Receiver's instance variables: sender: AioEventHandler>>initializeForExceptions:readEvents:writeEvents: pc: 32 stackp: 3 method: (BlockClosure>>#on:do: "a CompiledMethod(1660548)") closureOrNil: nil receiver: [closure] in AioEventHandler>>initializeForExceptions:readEvents:writeEvents:...etc... BlockClosure>>ensure: Receiver: [closure] in Context>>handleSignal: Arguments and temporary variables: aBlock: [closure] in Context>>handleSignal: complete: nil returnValue: nil Receiver's instance variables: outerContext: Context>>handleSignal: startpc: 174 numArgs: 0 Context>>handleSignal: Receiver: BlockClosure>>on:do: Arguments and temporary variables: exception: Warning: aio event forwarding not supported handlerActive: true val: nil Receiver's instance variables: sender: AioEventHandler>>initializeForExceptions:readEvents:writeEvents: pc: 32 stackp: 3 method: (BlockClosure>>#on:do: "a CompiledMethod(1660548)") closureOrNil: nil receiver: [closure] in AioEventHandler>>initializeForExceptions:readEvents:writeEvents:...etc... Warning(Exception)>>signal Receiver: Warning: aio event forwarding not supported Arguments and temporary variables: Receiver's instance variables: messageText: 'aio event forwarding not supported' tag: nil signalContext: Warning(Exception)>>signal handlerContext: BlockClosure>>on:do: outerContext: nil Warning(Exception)>>signal: Receiver: Warning: aio event forwarding not supported Arguments and temporary variables: signalerText: 'aio event forwarding not supported' Receiver's instance variables: messageText: 'aio event forwarding not supported' tag: nil signalContext: Warning(Exception)>>signal handlerContext: BlockClosure>>on:do: outerContext: nil Warning class>>signal: Receiver: Warning Arguments and temporary variables: signalerText: 'aio event forwarding not supported' Receiver's instance variables: superclass: Notification methodDict: a MethodDictionary(#defaultAction->(Warning>>#defaultAction "a CompiledMethod...etc... format: 65541 instanceVariables: nil organization: ('priv handling' defaultAction) subclasses: {Deprecation . ServiceCancelled . SAXWarning . UndeclaredVariableWarning...etc... name: #Warning classPool: nil sharedPools: nil environment: nil category: #'Kernel-Exceptions' suppressed: false resetOnStartUp: false AioEventHandler(Object)>>notify: Receiver: an AioEventHandler Arguments and temporary variables: aString: 'aio event forwarding not supported' Receiver's instance variables: dependents: nil semaphore: nil semaIndex: nil handlerProc: nil descriptor: nil AioEventHandler>>aioEnable:forSemaphore:externalObject: Receiver: an AioEventHandler Arguments and temporary variables: ioDescriptor: nil semaphoreIndex: 17 trueOrFalse: true Receiver's instance variables: dependents: nil semaphore: nil semaIndex: nil handlerProc: nil descriptor: nil [] in AioEventHandler>>initializeForExceptions:readEvents:writeEvents: Receiver: an AioEventHandler Arguments and temporary variables: <<error during printing> Receiver's instance variables: dependents: nil semaphore: nil semaIndex: nil handlerProc: nil descriptor: nil BlockClosure>>on:do: Receiver: [closure] in AioEventHandler>>initializeForExceptions:readEvents:writeEvents: Arguments and temporary variables: exceptionOrExceptionSet: Warning handlerAction: [closure] in AioEventHandler>>initializeForExceptions:readEvents:writeEvents:...etc... handlerActive: false Receiver's instance variables: outerContext: AioEventHandler>>initializeForExceptions:readEvents:writeEvents: startpc: 140 numArgs: 0 AioEventHandler>>initializeForExceptions:readEvents:writeEvents: Receiver: an AioEventHandler Arguments and temporary variables: exceptionEventFlag: true readEventFlag: true writeEventFlag: false Receiver's instance variables: dependents: nil semaphore: nil semaIndex: nil handlerProc: nil descriptor: nil AioEventHandler>>forFileStream:exceptions:readEvents:writeEvents: Receiver: an AioEventHandler Arguments and temporary variables: aFileStream: BufferedAsyncFileReadStream: 'pipeReader' exceptionEventFlag: true readEventFlag: true writeEventFlag: false Receiver's instance variables: dependents: nil semaphore: nil semaIndex: nil handlerProc: nil descriptor: nil AioEventHandler class>>forFileStream:exceptions:readEvents:writeEvents: Receiver: AioEventHandler Arguments and temporary variables: aFileStream: BufferedAsyncFileReadStream: 'pipeReader' exceptionEventFlag: true readEventFlag: true writeEventFlag: false Receiver's instance variables: superclass: Model methodDict: a MethodDictionary(#aioDisable:->(AioEventHandler>>#aioDisable: "a CompiledMethod...etc... format: 65541 instanceVariables: #('semaphore' 'semaIndex' 'handlerProc' 'descriptor') organization: ('aio event forwarding' aioDisable: aioEnable:forSemaphore:externalObject:...etc... subclasses: nil name: #AioEventHandler classPool: a Dictionary(#AioPluginPresent->true ) sharedPools: nil environment: Smalltalk category: #'OSProcess-AIO' BufferedAsyncFileReadStream(AsyncFileReadStream)>>enableEventHandling Receiver: BufferedAsyncFileReadStream: 'pipeReader' Arguments and temporary variables: Receiver's instance variables: BufferedAsyncFileReadStream(AsyncFileReadStream)>>initialize Receiver: BufferedAsyncFileReadStream: 'pipeReader' Arguments and temporary variables: Receiver's instance variables: BufferedAsyncFileReadStream class(AttachableFileStream class)>>name:attachTo:writable: Receiver: BufferedAsyncFileReadStream Arguments and temporary variables: aSymbolOrString: 'pipeReader' anIOHandle: #[93 179 159 128 0 0 0 0 48 162 123 2 0 0 0 0 0 0 0 0 0 0 0 0] readWriteFlag: false Receiver's instance variables: superclass: AsyncFileReadStream methodDict: a MethodDictionary(#actAsExecutor->(BufferedAsyncFileReadStream>>#a...etc... format: 65552 instanceVariables: #('nonBlockingMode' 'readBuffer' 'readSyncSemaphore' 'dataAv...etc... organization: ('finalization' actAsExecutor) ('read ahead buffer' appendAllToBuffer:...etc... subclasses: nil name: #BufferedAsyncFileReadStream classPool: nil sharedPools: nil environment: Smalltalk category: #'OSProcess-Base' --- The full stack --- AioEventHandler(Object)>>notify: [] in AioEventHandler>>initializeForExceptions:readEvents:writeEvents: BlockClosure>>cull: [] in Context>>handleSignal: BlockClosure>>ensure: Context>>handleSignal: Warning(Exception)>>signal Warning(Exception)>>signal: Warning class>>signal: AioEventHandler(Object)>>notify: AioEventHandler>>aioEnable:forSemaphore:externalObject: [] in AioEventHandler>>initializeForExceptions:readEvents:writeEvents: BlockClosure>>on:do: AioEventHandler>>initializeForExceptions:readEvents:writeEvents: AioEventHandler>>forFileStream:exceptions:readEvents:writeEvents: AioEventHandler class>>forFileStream:exceptions:readEvents:writeEvents: BufferedAsyncFileReadStream(AsyncFileReadStream)>>enableEventHandling BufferedAsyncFileReadStream(AsyncFileReadStream)>>initialize BufferedAsyncFileReadStream class(AttachableFileStream class)>>name:attachTo:writable: - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - AttachableFileStream>>asBufferedAsyncFileReadStream ExternalPipe>>setBufferedReader PipeableOSProcess(PipeJunction)>>enableOutputStreamTriggers PipeableOSProcess>>enableEventTriggers PipeableOSProcess(PipeJunction)>>prepareOutputForInternalReader PipeableOSProcess>>prepareOutputForInternalReader PipeableOSProcess(PipeJunction)>>prepareOutputFor: ProxyPipeline>>fromString:shell: ProxyPipeline class>>fromString:shell: CommandShell>>pipelineCollectionFrom: CommandShell>>evaluateOrMakePipelinesFrom: CommandShell>>command:echo: [] in CommandShell>>processCommand:echo: [] in BlockClosure>>newProcess

3 7

primitive 87 resume
by Florin Mateoc 21 May '21

21 May '21

Hi all, In the method Process>>resume, which invokes primitive 87, the comment states that the primitive should fail if the process is already waiting in a queue. But the implementation does not do that. If we execute (inspect) the snippet: | evaluationProcess result delay testProcess | delay := Delay forMilliseconds: 50. testProcess := Processor activeProcess. evaluationProcess := [ delay unschedule. result := testProcess isBlocked. testProcess resume] fork. delay wait. result it returns true (and no primitive failed error occurs), which means we successfully resumed a blocked (waiting on a semaphore) process. I also checked in VMMaker and indeed, the code does not check if the process is on a queue or not before just putting it on a runnable list. What do you think? Should the comment be changed or the implementation? Florin

2 4

VM Maker: VMMaker.oscog-eem.2961.mcz
by commits＠source.squeak.org 21 May '21

21 May '21

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.2961.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2961 Author: eem Time: 20 May 2021, 3:54:32.723618 pm UUID: fa08f5a0-f381-4cee-9c68-6a7610fd2728 Ancestors: VMMaker.oscog-eem.2960 Tweak primitiveResume to get better Slang inlining. =============== Diff against VMMaker.oscog-eem.2960 =============== Item was changed: ----- Method: CoInterpreterPrimitives>>primitiveResume (in category 'process primitives') ----- primitiveResume "Put this process on the scheduler's lists thus allowing it to proceed next time there is a chance for processes of it's priority level. It must go to the back of its run queue so as not to preempt any already running processes at this level. If the process's priority is higher than the current process, preempt the current process." + | proc ctxt inInterpreter | - | proc inInterpreter | proc := self stackTop. "rcvr" + ctxt := objectMemory followField: SuspendedContextIndex ofObject: proc. "written this way to get better Slang inlining" + (objectMemory isContext: ctxt) ifFalse: + [^self primitiveFailFor: PrimErrBadReceiver]. - (objectMemory isContext: (objectMemory followField: SuspendedContextIndex ofObject: proc)) ifFalse: - [^self primitiveFail]. "We're about to switch process, either to an interpreted frame or a machine code frame. To know whether to return or enter machine code we have to know from whence we came. We could have come from the interpreter, either directly or via a machine code primitive. We could have come from machine code. The instructionPointer tells us where from:" inInterpreter := instructionPointer >= objectMemory startOfMemory. (self resume: proc preemptedYieldingIf: preemptionYields from: CSResume) ifTrue: [self forProcessPrimitiveReturnToExecutivePostContextSwitch: inInterpreter] "Personally I would like to check MyList, which should not be one of the elements of the scheduler lists. But there are awful race conditions in things like should:notTakeMoreThan: that mean we can't. eem 9/27/2010 23:08, updated eem 5/20/2021 15:36. e.g. + | proc field | - | proc | proc := self stackTop. ''rcvr'' ''We only have to check for myList being nil. If it is nil then this is either the active process or a process suspended with primitiveSuspend (and if it is the activeProcess suspendedContext will be nil and the isContext: test will fail). If it is not nil then either the process is waiting on some semaphore-like list or on one of the scheduler's lists. If it is on some semaphore-like list it should not resume. If it is on one of the scheduler's lists it is runnable (already resumed).'' + field := objectMemory followField: MyListIndex ofObject: proc. + objectMemory nilObject = field ifFalse: - objectMemory nilObject = (objectMemory followField: MyListIndex ofObject: proc) ifFalse: [^self primitiveFailFor: PrimErrInappropriate]. + field := objectMemory followField: SuspendedContextIndex ofObject: proc. + (objectMemory isContext: field) ifFalse: - (objectMemory isContext: (objectMemory followField: SuspendedContextIndex ofObject: proc)) ifFalse: [^self primitiveFailFor: PrimErrBadReceiver]. ''We're about to switch process, either to an interpreted frame or a machine code frame. To know whether to return or enter machine code we have to know from whence we came. We could have come from the interpreter, either directly or via a machine code primitive. We could have come from machine code. The instructionPointer tells us where from:'' inInterpreter := instructionPointer >= objectMemory startOfMemory. (self resume: proc preemptedYieldingIf: preemptionYields from: CSResume) ifTrue: [self forProcessPrimitiveReturnToExecutivePostContextSwitch: inInterpreter]"! Item was changed: ----- Method: InterpreterPrimitives>>primitiveResume (in category 'process primitives') ----- primitiveResume "Put this process on the scheduler's lists thus allowing it to proceed next time there is a chance for processes of it's priority level. It must go to the back of its run queue so as not to preempt any already running processes at this level. If the process's priority is higher than the current process, preempt the current process." + | proc ctxt | - | proc | proc := self stackTop. "rcvr" + ctxt := objectMemory followField: SuspendedContextIndex ofObject: proc. "written this way to get better Slang inlining" + (objectMemory isContext: ctxt) ifFalse: + [^self primitiveFailFor: PrimErrBadReceiver]. - (objectMemory isContext: (objectMemory followField: SuspendedContextIndex ofObject: proc)) ifFalse: - [^self primitiveFail]. self resume: proc preemptedYieldingIf: preemptionYields "Personally I would like to check MyList, which should not be one of the elements of the scheduler lists. But there are awful race conditions in things like should:notTakeMoreThan: that mean we can't. eem 9/27/2010 23:08, updated eem 5/20/2021 15:36. e.g. + | proc field | - | proc | proc := self stackTop. ''rcvr'' ''We only have to check for myList being nil. If it is nil then this is either the active process or a process suspended with primitiveSuspend (and if it is the activeProcess suspendedContext will be nil and the isContext: test will fail). If it is not nil then either the process is waiting on some semaphore-like list or on one of the scheduler's lists. If it is on some semaphore-like list it should not resume. If it is on one of the scheduler's lists it is runnable (already resumed).'' + field := objectMemory followField: MyListIndex ofObject: proc. + objectMemory nilObject = field ifFalse: - objectMemory nilObject = (objectMemory followField: MyListIndex ofObject: proc) ifFalse: [^self primitiveFailFor: PrimErrInappropriate]. + field := objectMemory followField: SuspendedContextIndex ofObject: proc. + (objectMemory isContext: field) ifFalse: - (objectMemory isContext: (objectMemory followField: SuspendedContextIndex ofObject: proc)) ifFalse: [^self primitiveFailFor: PrimErrBadReceiver]. self resume: proc preemptedYieldingIf: preemptionYields"!

1 0

VM Maker: VMMaker.oscog-eem.2960.mcz
by commits＠source.squeak.org 21 May '21

21 May '21

Eliot Miranda uploaded a new version of VMMaker to project VM Maker: http://source.squeak.org/VMMaker/VMMaker.oscog-eem.2960.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2960 Author: eem Time: 20 May 2021, 3:39:38.287864 pm UUID: 9f7e6567-73d5-4991-83e1-e1958fd55611 Ancestors: VMMaker.oscog-eem.2958 InterpreterPrimitives>>primitiveResume should also follow the SuspendedContextIndex field of a process on Spur, as does the CoInterpreterPrimitives version. Update the comment re not checking myList. See VMMaker.oscog-eem.2959 in VMMakerInbox for a version that does check and is being considered for inclusion. =============== Diff against VMMaker.oscog-eem.2958 =============== Item was changed: ----- Method: CoInterpreterPrimitives>>primitiveResume (in category 'process primitives') ----- primitiveResume "Put this process on the scheduler's lists thus allowing it to proceed next time there is a chance for processes of it's priority level. It must go to the back of its run queue so as not to preempt any already running processes at this level. If the process's priority is higher than the current process, preempt the current process." | proc inInterpreter | proc := self stackTop. "rcvr" (objectMemory isContext: (objectMemory followField: SuspendedContextIndex ofObject: proc)) ifFalse: [^self primitiveFail]. "We're about to switch process, either to an interpreted frame or a machine code frame. To know whether to return or enter machine code we have to know from whence we came. We could have come from the interpreter, either directly or via a machine code primitive. We could have come from machine code. The instructionPointer tells us where from:" inInterpreter := instructionPointer >= objectMemory startOfMemory. (self resume: proc preemptedYieldingIf: preemptionYields from: CSResume) ifTrue: [self forProcessPrimitiveReturnToExecutivePostContextSwitch: inInterpreter] "Personally I would like to check MyList, which should not be one of the elements of the scheduler lists. But there are awful race conditions in things like should:notTakeMoreThan: that mean we can't. + eem 9/27/2010 23:08, updated eem 5/20/2021 15:36. e.g. - eem 9/27/2010 23:08. e.g. + | proc | + proc := self stackTop. ''rcvr'' + ''We only have to check for myList being nil. If it is nil then this is either the active process or + a process suspended with primitiveSuspend (and if it is the activeProcess suspendedContext will + be nil and the isContext: test will fail). If it is not nil then either the process is waiting on some + semaphore-like list or on one of the scheduler's lists. If it is on some semaphore-like list it should + not resume. If it is on one of the scheduler's lists it is runnable (already resumed).'' + objectMemory nilObject = (objectMemory followField: MyListIndex ofObject: proc) ifFalse: + [^self primitiveFailFor: PrimErrInappropriate]. + (objectMemory isContext: (objectMemory followField: SuspendedContextIndex ofObject: proc)) ifFalse: + [^self primitiveFailFor: PrimErrBadReceiver]. - | proc myList classLinkedList | - proc := self stackTop. - myList := objectMemory fetchPointer: MyListIndex ofObject: proc. - classLinkedList := self superclassOf: (objectMemory splObj: ClassSemaphore). - ((self fetchClassOfNonInt: myList) ~= classLinkedList - and: [objectMemory isContext: (objectMemory fetchPointer: SuspendedContextIndex ofObject: proc)]) ifFalse: - [^self primitiveFail]. ''We're about to switch process, either to an interpreted frame or a machine code frame. To know whether to return or enter machine code we have to know from whence we came. We could have come from the interpreter, either directly or via a machine code primitive. We could have come from machine code. The instructionPointer tells us where from:'' inInterpreter := instructionPointer >= objectMemory startOfMemory. + (self resume: proc preemptedYieldingIf: preemptionYields from: CSResume) ifTrue: - (self resume: proc preemptedYieldingIf: preemptionYields from: CSResume) ifTrue: [self forProcessPrimitiveReturnToExecutivePostContextSwitch: inInterpreter]"! Item was changed: ----- Method: InterpreterPrimitives>>primitiveResume (in category 'process primitives') ----- primitiveResume "Put this process on the scheduler's lists thus allowing it to proceed next time there is a chance for processes of it's priority level. It must go to the back of its run queue so as not to preempt any already running processes at this level. If the process's priority is higher than the current process, preempt the current process." | proc | proc := self stackTop. "rcvr" + (objectMemory isContext: (objectMemory followField: SuspendedContextIndex ofObject: proc)) ifFalse: - (objectMemory isContext: (objectMemory fetchPointer: SuspendedContextIndex ofObject: proc)) ifFalse: [^self primitiveFail]. self resume: proc preemptedYieldingIf: preemptionYields "Personally I would like to check MyList, which should not be one of the elements of the scheduler lists. But there are awful race conditions in things like should:notTakeMoreThan: that mean we can't. + eem 9/27/2010 23:08, updated eem 5/20/2021 15:36. e.g. - eem 9/27/2010 23:08. e.g. + | proc | + proc := self stackTop. ''rcvr'' + ''We only have to check for myList being nil. If it is nil then this is either the active process or + a process suspended with primitiveSuspend (and if it is the activeProcess suspendedContext will + be nil and the isContext: test will fail). If it is not nil then either the process is waiting on some + semaphore-like list or on one of the scheduler's lists. If it is on some semaphore-like list it should + not resume. If it is on one of the scheduler's lists it is runnable (already resumed).'' + objectMemory nilObject = (objectMemory followField: MyListIndex ofObject: proc) ifFalse: + [^self primitiveFailFor: PrimErrInappropriate]. + (objectMemory isContext: (objectMemory followField: SuspendedContextIndex ofObject: proc)) ifFalse: + [^self primitiveFailFor: PrimErrBadReceiver]. - | proc myList classLinkedList | - proc := self stackTop. - myList := objectMemory fetchPointer: MyListIndex ofObject: proc. - classLinkedList := self superclassOf: (objectMemory splObj: ClassSemaphore). - ((self fetchClassOfNonInt: myList) ~= classLinkedList - and: [objectMemory isContext: (objectMemory fetchPointer: SuspendedContextIndex ofObject: proc)]) ifFalse: - [^self primitiveFail]. self resume: proc preemptedYieldingIf: preemptionYields"!

1 0

VM Maker Inbox: VMMaker.oscog-eem.2959.mcz
by commits＠source.squeak.org 21 May '21

21 May '21

Eliot Miranda uploaded a new version of VMMaker to project VM Maker Inbox: http://source.squeak.org/VMMakerInbox/VMMaker.oscog-eem.2959.mcz ==================== Summary ==================== Name: VMMaker.oscog-eem.2959 Author: eem Time: 20 May 2021, 3:32:35.376631 pm UUID: bcb1f4a6-116d-4a10-ae3f-fcd449535d7d Ancestors: VMMaker.oscog-eem.2958 Candidate implementations of primitiveResume that check for myList and fail if it is not nil. See the comment in InterpreterPrimitives>>primitiveResume "Personally I would like to check MyList, which should not be one of the elements of the scheduler lists. But there are awful race conditions in things like should:notTakeMoreThan: that mean we can't. eem 9/27/2010 23:08." =============== Diff against VMMaker.oscog-eem.2958 =============== Item was changed: ----- Method: CoInterpreterPrimitives>>primitiveResume (in category 'process primitives') ----- primitiveResume "Put this process on the scheduler's lists thus allowing it to proceed next time there is a chance for processes of it's priority level. It must go to the back of its run queue so as not to preempt any already running processes at this level. If the process's priority is higher than the current process, preempt the current process." | proc inInterpreter | proc := self stackTop. "rcvr" + "We only have to check for myList being nil. If it is nil then this is either the active process or + a process suspended with primitiveSuspend (and if it is the activeProcess suspendedContext will + be nil and the isContext: test will fail). If it is not nil then either the process is waiting on some + semaphore-like list or on one of the scheduler's lists. If it is on some semaphore-like list it should + not resume. If it is on one of the scheduler's lists it is runnable (already resumed)." + objectMemory nilObject = (objectMemory followField: MyListIndex ofObject: proc) ifFalse: + [^self primitiveFailFor: PrimErrInappropriate]. (objectMemory isContext: (objectMemory followField: SuspendedContextIndex ofObject: proc)) ifFalse: + [^self primitiveFailFor: PrimErrBadReceiver]. - [^self primitiveFail]. "We're about to switch process, either to an interpreted frame or a machine code frame. To know whether to return or enter machine code we have to know from whence we came. We could have come from the interpreter, either directly or via a machine code primitive. We could have come from machine code. The instructionPointer tells us where from:" inInterpreter := instructionPointer >= objectMemory startOfMemory. (self resume: proc preemptedYieldingIf: preemptionYields from: CSResume) ifTrue: [self forProcessPrimitiveReturnToExecutivePostContextSwitch: inInterpreter] "Personally I would like to check MyList, which should not be one of the elements of the scheduler lists. But there are awful race conditions in things like should:notTakeMoreThan: that mean we can't. eem 9/27/2010 23:08. e.g. | proc myList classLinkedList | proc := self stackTop. myList := objectMemory fetchPointer: MyListIndex ofObject: proc. classLinkedList := self superclassOf: (objectMemory splObj: ClassSemaphore). ((self fetchClassOfNonInt: myList) ~= classLinkedList and: [objectMemory isContext: (objectMemory fetchPointer: SuspendedContextIndex ofObject: proc)]) ifFalse: [^self primitiveFail]. ''We're about to switch process, either to an interpreted frame or a machine code frame. To know whether to return or enter machine code we have to know from whence we came. We could have come from the interpreter, either directly or via a machine code primitive. We could have come from machine code. The instructionPointer tells us where from:'' inInterpreter := instructionPointer >= objectMemory startOfMemory. (self resume: proc preemptedYieldingIf: preemptionYields from: CSResume) ifTrue: [self forProcessPrimitiveReturnToExecutivePostContextSwitch: inInterpreter]"! Item was changed: ----- Method: InterpreterPrimitives>>primitiveResume (in category 'process primitives') ----- primitiveResume "Put this process on the scheduler's lists thus allowing it to proceed next time there is a chance for processes of it's priority level. It must go to the back of its run queue so as not to preempt any already running processes at this level. If the process's priority is higher than the current process, preempt the current process." | proc | proc := self stackTop. "rcvr" + "We only have to check for myList being nil. If it is nil then this is either the active process or + a process suspended with primitiveSuspend (and if it is the activeProcess suspendedContext will + be nil and the isContext: test will fail). If it is not nil then either the process is waiting on some + semaphore-like list or on one of the scheduler's lists. If it is on some semaphore-like list it should + not resume. If it is on one of the scheduler's lists it is runnable (already resumed)." + objectMemory nilObject = (objectMemory followField: MyListIndex ofObject: proc) ifFalse: + [^self primitiveFailFor: PrimErrInappropriate]. + (objectMemory isContext: (objectMemory followField: SuspendedContextIndex ofObject: proc)) ifFalse: + [^self primitiveFailFor: PrimErrBadReceiver]. - (objectMemory isContext: (objectMemory fetchPointer: SuspendedContextIndex ofObject: proc)) ifFalse: - [^self primitiveFail]. self resume: proc preemptedYieldingIf: preemptionYields "Personally I would like to check MyList, which should not be one of the elements of the scheduler lists. But there are awful race conditions in things like should:notTakeMoreThan: that mean we can't. eem 9/27/2010 23:08. e.g. | proc myList classLinkedList | proc := self stackTop. myList := objectMemory fetchPointer: MyListIndex ofObject: proc. classLinkedList := self superclassOf: (objectMemory splObj: ClassSemaphore). ((self fetchClassOfNonInt: myList) ~= classLinkedList and: [objectMemory isContext: (objectMemory fetchPointer: SuspendedContextIndex ofObject: proc)]) ifFalse: [^self primitiveFail]. self resume: proc preemptedYieldingIf: preemptionYields"!

1 0

[OpenSmalltalk/opensmalltalk-vm] config.h:40:2: error: This file was included too late. Please make sure it is included earlier (#567)
by asarch 21 May '21

21 May '21

This is what happens in Debian 10 Buster for AMD64: [compilacion.log](https://github.com/OpenSmalltalk/opensmalltalk-vm/files/65… -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/567

2 2

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

Vm-dev May 2021